Note: Descriptions are shown in the official language in which they were submitted.
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
ADAPTIVE BILATERAL MATCHING FOR DECODER SIDE MOTION
VECTOR REFINEMENT
FIELD
[0001] The present disclosure generally relates to video encoding and
decoding. For example,
aspects of the present disclosure include improving video coding techniques
related to decoder-
side motion vector refinement (DMVR) using bilateral matching.
BACKGROUND
[0002] Digital video capabilities can be incorporated into a wide range of
devices, including
digital televisions, digital direct broadcast systems, wireless broadcast
systems, personal digital
assistants (PDAs), laptop or desktop computers, tablet computers, e-book
readers, digital cameras,
digital recording devices, digital media players, video gaming devices, video
game consoles,
cellular or satellite radio telephones, so-called "smart phones," video
teleconferencing devices,
video streaming devices, and the like. Such devices allow video data to be
processed and output
for consumption. Digital video data includes large amounts of data to meet the
demands of
consumers and video providers. For example, consumers of video data desire
video of the utmost
quality, with high fidelity, resolutions, frame rates, and the like. As a
result, the large amount of
video data that is required to meet these demands places a burden on
communication networks and
devices that process and store the video data.
[0003] Digital video devices can implement video coding techniques to compress
video data.
Video coding is performed according to one or more video coding standards or
formats. For
example, video coding standards or formats include versatile video coding
(VVC), high-efficiency
video coding (HEVC), advanced video coding (AVC), MPEG-2 Part 2 coding (MPEG
stands for
moving picture experts group), among others, as well as proprietary video
codecs/formats such as
AOMedia Video 1 (AV1) that was developed by the Alliance for Open Media. Video
coding
generally utilizes prediction methods (e.g., inter prediction, intra
prediction, or the like) that take
advantage of redundancy present in video images or sequences. A goal of video
coding techniques
is to compress video data into a form that uses a lower bit rate, while
avoiding or minimizing
1
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
degradations to video quality. With ever-evolving video services becoming
available, coding
techniques with better coding efficiency are needed.
BRIEF SUM MARY
[0004] In some examples, systems and techniques are described for decoder-side
motion vector
refinement (DMVR) using adaptive bilateral matching. According to at least one
illustrative
example, an apparatus for processing video data is provided that includes at
least one memory
(e.g., configured to store data, such as video data, etc.) and at least one
processor (e.g.,
implemented in circuitry) coupled to the at least one memory. The at least one
processor is
configured to and can: obtain one or more reference pictures for a current
picture; identify a first
motion vector and a second motion vector for a merge mode candidate; determine
a selected
motion vector search strategy for the merge mode candidate from a plurality of
motion vector
search strategies; determine, using the selected motion vector search
strategy, one or more refined
motion vectors based on at least one of the first motion vector or the second
motion vector and the
one or more reference pictures; and process the merge mode candidate using the
one or more
refined motion vectors.
[0005] In another example, a method for processing video data is provided. The
method
includes: obtaining one or more reference pictures for a current picture;
identifying a first motion
vector and a second motion vector for a merge mode candidate; determining a
selected motion
vector search strategy for the merge mode candidate from a plurality of motion
vector search
strategies; determining, using the selected motion vector search strategy, one
or more refined
motion vectors based on at least one of the first motion vector or the second
motion vector and the
one or more reference pictures; and processing the merge mode candidate using
the one or more
refined motion vectors.
[0006] In another example, a non-transitory computer-readable medium is
provided that has
stored thereon instructions that, when executed by one or more processors,
cause the one or more
processors to: obtain one or more reference pictures for a current picture;
identify a first motion
vector and a second motion vector for a merge mode candidate; determine a
selected motion vector
search strategy for the merge mode candidate from a plurality of motion vector
search strategies;
determine, using the selected motion vector search strategy, one or more
refined motion vectors
2
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
based on at least one of the first motion vector or the second motion vector
and the one or more
reference pictures; and process the merge mode candidate using the one or more
refined motion
vectors.
[0007] In another example, an apparatus for processing video data is provided.
The apparatus
includes: means for obtaining one or more reference pictures for a current
picture; means for
identifying a first motion vector and a second motion vector for a merge mode
candidate; means
for determining a selected motion vector search strategy for the merge mode
candidate from a
plurality of motion vector search strategies; means for determining, using the
selected motion
vector search strategy, one or more refined motion vectors based on at least
one of the first motion
vector or the second motion vector and the one or more reference pictures; and
means for
processing the merge mode candidate using the one or more refined motion
vectors.
[0008] This summary is not intended to identify key or essential features of
the claimed subject
matter, nor is it intended to be used in isolation to determine the scope of
the claimed subject
matter. The subject matter should be understood by reference to appropriate
portions of the entire
specification of this patent, any or all drawings, and each claim.
[0009] The foregoing, together with other features and aspects, will become
more apparent upon
referring to the following specification, claims, and accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
[0010] Illustrative aspects of the present application are described in detail
below with reference
to the following figures:
[0011] FIG. 1 is a block diagram illustrating an example of an encoding device
and a decoding
device, in accordance with some examples of the disclosure;
[0012] FIG. 2A is a conceptual diagram illustrating example spatial
neighboring motion vector
candidates for a merge mode, in accordance with some examples of the
disclosure;
[0013] FIG. 2B is a conceptual diagram illustrating example spatial
neighboring motion vector
candidates for an advanced motion vector prediction (AMVP) mode, in accordance
with some
examples of the disclosure;
3
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0014] FIG. 3A is a conceptual diagram illustrating an example temporal motion
vector predictor
(TMVP) candidate, in accordance with some examples of the disclosure;
[0015] FIG. 3B is a conceptual diagram illustrating an example of motion
vector scaling, in
accordance with some examples of the disclosure;
[0016] FIG. 4A is a conceptual diagram illustrating an example of neighboring
samples of a
current coding unit used for estimating motion compensation parameters for the
current coding
unit, in accordance with some examples of the disclosure;
[0017] FIG. 4B is a conceptual diagram illustrating an example of neighboring
samples of a
reference block used for estimating motion compensation parameters for a
current coding unit, in
accordance with some examples of the disclosure;
[0018] FIG. 5 illustrates locations of spatial merge candidates for use in
processing a block in
accordance with some examples of the disclosure;
[0019] FIG. 6 illustrates aspects of motion vector scaling for temporal merge
candidates for use
in processing a block in accordance with some examples of the disclosure;
[0020] FIG. 7 illustrates aspects of temporal merge candidates for use in
processing a block in
accordance with some examples of the disclosure;
[0021] FIG. 8 illustrates aspects of bilateral matching in accordance with
some examples of the
disclosure;
[0022] FIG. 9 illustrates aspects of bi-directional optical flow (BDOF) in
accordance with some
examples of the disclosure;
[0023] FIG. 10 illustrates search area regions in accordance with some
examples of the
disclosure;
[0024] FIG. 11 is a flowchart illustrating an example process for decoder side
motion vector
refinement with adaptive bilateral matching, in accordance with some examples
of the disclosure;
4
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0025] FIG. 12 is a block diagram illustrating an example video encoding
device, in accordance
with some examples of the disclosure; and
[0026] FIG. 13 is a block diagram illustrating an example video decoding
device, in accordance
with some examples of the disclosure.
DETAILED DESCRIPTION
[0027] Certain aspects and aspects of this disclosure are provided below. Some
of these aspects
and aspects may be applied independently and some of them may be applied in
combination as
would be apparent to those of skill in the art. In the following description,
for the purposes of
explanation, specific details are set forth in order to provide a thorough
understanding of aspects
of the application. However, it will be apparent that various aspects may be
practiced without these
specific details. The figures and description are not intended to be
restrictive.
[0028] The ensuing description provides exemplary aspects only, and is not
intended to limit the
scope, applicability, or configuration of the disclosure. Rather, the ensuing
description of the
exemplary aspects will provide those skilled in the art with an enabling
description for
implementing an exemplary aspect. It should be understood that various changes
may be made in
the function and arrangement of elements without departing from the spirit and
scope of the
application as set forth in the appended claims.
[0029] Video coding devices (e.g., encoding devices, decoding devices, or
combined encoding-
decoding devices) implement video compression techniques to encode and/or
decode video data
efficiently. Video compression techniques may include applying different
prediction modes,
including spatial prediction (e.g., intra-frame prediction or intra-
prediction), temporal prediction
(e.g., inter-frame prediction or inter-prediction), inter-layer prediction
(across different layers of
video data), and/or other prediction techniques to reduce or remove redundancy
inherent in video
sequences. A video encoder can partition each picture of an original video
sequence into
rectangular regions referred to as video blocks or coding units (described in
greater detail below).
These video blocks may be encoded using a particular prediction mode.
[0030] Video blocks may be divided in one or more ways into one or more groups
of smaller
blocks. Blocks can include coding tree blocks, prediction blocks, transform
blocks, or other
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
suitable blocks. References generally to a "block," unless otherwise
specified, may refer to such
video blocks (e.g., coding tree blocks, coding blocks, prediction blocks,
transform blocks, or other
appropriate blocks or sub-blocks, as would be understood by one of ordinary
skill). Further, each
of these blocks may also interchangeably be referred to herein as "units"
(e.g., coding tree unit
(CTU), coding unit, prediction unit (PU), transform unit (TU), or the like).
In some cases, a unit
may indicate a coding logical unit that is encoded in a bitstream, while a
block may indicate a
portion of video frame buffer a process is target to.
[0031] For inter-prediction modes, a video encoder can search for a block
similar to the block
being encoded in a frame (or picture) located in another temporal location,
referred to as a
reference frame or a reference picture. The video encoder may restrict the
search to a certain spatial
displacement from the block to be encoded. A best match may be located using a
two-dimensional
(2D) motion vector that includes a horizontal displacement component and a
vertical displacement
component. For intra-prediction modes, a video encoder may form the predicted
block using
spatial prediction techniques based on data from previously encoded
neighboring blocks within
the same picture.
[0032] The video encoder may determine a prediction error. For example, the
prediction can be
determined as the difference between the pixel values in the block being
encoded and the predicted
block. The prediction error can also be referred to as the residual. The video
encoder may also
apply a transform to the prediction error (e.g., a discrete cosine transform
(DCT) or other suitable
transform) to generate transform coefficients. After transformation, the video
encoder may
quantize the transform coefficients. The quantized transform coefficients and
motion vectors may
be represented using syntax elements and, along with control information, form
a coded
representation of a video sequence. In some instances, the video encoder may
entropy code syntax
elements, thereby further reducing the number of bits needed for their
representation.
[0033] A video decoder may construct, using the syntax elements and control
information
discussed above, predictive data (e.g., a predictive block) for decoding a
current frame. For
example, the video decoder may add the predicted block and the compressed
prediction error. The
video decoder may determine the compressed prediction error by weighting the
transform basis
6
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
functions using the quantized coefficients. The difference between the
reconstructed frame and the
original frame is called reconstruction error.
[0034] Systems, apparatuses, processes (also referred to as methods), and
computer-readable
media (collectively referred to herein as "systems and techniques") are
described herein for
increasing the accuracy of one or more motion vectors that can be used by a
video coding device
(e.g., a video decoder or decoding device) when performing a prediction
technique (e.g., an inter-
prediction mode). For example, the systems and techniques can perform
bilateral matching for
decoder-side motion vector refinement (DMVR). Bilateral matching is a
technique that refines a
pair of two initial motion vectors. Such refinement can occur with a search
around the pair of initial
motion vectors to derive updated motion vectors that minimize a block matching
cost. The block
matching cost can be generated in a variety of ways, including using a sum of
absolute difference
(SAD) criteria, a sum of absolute transformed difference (SATD) criteria, a
sum of square error
(SSE) criteria, or other such criteria. Aspects described herein can increase
the accuracy of motion
vectors of a bi-prediction merge candidate, resulting in improved video
quality and associated
improved performance of devices operating in accordance with the aspects
described herein.
[0035] In some aspects, the systems and techniques can be used to perform
adaptive bilateral
matching for DMVR. For example, the systems and techniques can perform
bilateral matching
using different search strategies and/or search parameters for different coded
blocks. As will be
explained in greater depth below, the adaptive bilateral matching for DMVR can
be based on a
selected search strategy that is determined or signaled for a given block. The
selected search
strategy can include one or more constraints for a bilateral matching search
process. In some
examples, the selected search strategy can additionally, or alternatively,
include one or more
constraints for a first motion vector difference and/or a second motion
difference. In some
examples, the selected search strategy can include one or more constraints
between a first motion
vector difference and a second motion vector difference.
[0036] In some aspects, a constraint is selected for motion vectors to be
refined. The constraint
can be a mirroring constraint, a zero constraint for a first vector, a zero
constraint for a second
vector, or other type of constraint. In some cases, the constraint is applied
to a merge mode coded
block within a merge candidate that satisfies one or more DMVR conditions. One
or more
7
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
constraints may then be used with one or more search strategies to identify
candidates and select
refined motion vectors.
[0037] In some aspects, different search strategies are used. The search
strategies can be grouped
into multiple subsets, with each subset including one or more search
strategies. In some cases, a
decoder can use a syntax element to determine a selected subset. For example,
an encoder can
include the syntax element in a bitstream. In such an example, the decoder may
receive the
bitstream and decode the syntax element from the bitstream. The decoder can
use the syntax
element to determine the selected subset and any associated constraint(s) for
a given block or
blocks of video data included in the bitstream. Using the selected subset and
any constraint(s)
associated with the subset, the decoder can process motion vectors (e.g., two
motion vectors of a
bi-prediction merge candidate) to identify refined motion vectors. In one
illustrative aspect, an
adaptive bilateral mode is provided, where a coding device signals selected
motion information
candidates that satisfy relevant DMVR conditions (e.g., with signaling
structures as part of a new
adaptive bilateral mode).
[0038] Using the above-noted search strategies and associated constraint(s)
can provide
improvements to decoder side motion vector refinement, such as by providing
adaptive bilateral
motion vector refinement using selectable search algorithms and associated
constraints. Such
improvements in decoder side motion vector refinement can be used with various
video codecs,
such as enhanced compression model (ECM) implementations. Examples described
herein include
an implementation applied to multi-pass DMVR to improve ECM systems that
operate according
to one or more video coding standards. The techniques described herein can be
implemented using
one or more coding devices, including one or more encoding devices, decoding
devices, or
combined encoding-decoding devices. The coding devices can be implemented by
one or more of
a player device, such as a mobile device, extended reality (XR) device, a
vehicle or computing
system of a vehicle, a server device or system (e.g., a distributed server
system including multiple
servers, a single server device or system, etc.), or other device or system.
[0039] The systems and techniques described herein can be applied to any
existing video codecs,
any video codecs that are in development, and/or any future video coding
standards, including but
not limited to High Efficiency Video Coding (HEVC), Advanced Video Coding
(AVC), Versatile
8
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
Video Coding (VVC), VP9, the AOMedia Video 1 (AV1) format/codec, and/or other
video coding
standard in existence, in development, or to be developed. The systems and
techniques described
herein can improve the operation of communication systems and devices in a
system by improving
the performance of video data transfer by devices with improved compression
and associated
improved video quality based on improved motion vector selection from adaptive
bilateral
matching as described herein.
[0040] FIG. 1 is a block diagram illustrating an example of a system 100
including an encoding
device 104 and a decoding device 112. The encoding device 104 may be part of a
source device,
and the decoding device 112 may be part of a receiving device. The source
device and/or the
receiving device may include an electronic device, such as a mobile or
stationary telephone handset
(e.g., smartphone, cellular telephone, or the like), a desktop computer, a
laptop or notebook
computer, a tablet computer, a set-top box, a television, a camera, a display
device, a digital media
player, a video gaming console, a video streaming device, an Internet Protocol
(IP) camera, or any
other suitable electronic device. In some examples, the source device and the
receiving device may
include one or more wireless transceivers for wireless communications. The
coding techniques
described herein are applicable to video coding in various multimedia
applications, including
streaming video transmissions (e.g., over the Internet), television broadcasts
or transmissions,
encoding of digital video for storage on a data storage medium, decoding of
digital video stored
on a data storage medium, or other applications. As used herein, the term
coding can refer to
encoding and/or decoding. In some examples, system 100 can support one-way or
two-way video
transmission to support applications such as video conferencing, video
streaming, video playback,
video broadcasting, gaming, and/or video telephony.
[0041] The encoding device 104 (or encoder) can be used to encode video data
using a video
coding standard, format, codec, or protocol to generate an encoded video
bitstream. Examples of
video coding standards and formats/codecs include ITU-T H.261, ISO/IEC MPEG-1
Visual, ITU-
T H.262 or ISO/IEC MPEG-2 Visual, ITU-T H.263, ISO/IEC MPEG-4 Visual, ITU-T
H.264 (also
known as ISO/IEC MPEG-4 AVC), including its Scalable Video Coding (SVC) and
Multiview
Video Coding (MVC) extensions, High Efficiency Video Coding (HEVC) or ITU-T
H.265, and
Versatile Video Coding (VVC) or ITU-T H.266. Various extensions to HEVC deal
with multi-
layer video coding exist, including the range and screen content coding
extensions, 3D video
9
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
coding (3D-HEVC) and multiview extensions (MV-HEVC) and scalable extension
(SHVC). The
HEVC and its extensions have been developed by the Joint Collaboration Team on
Video Coding
(JCT-VC) as well as Joint Collaboration Team on 3D Video Coding Extension
Development (JCT-
3V) of ITU-T Video Coding Experts Group (VCEG) and ISO/IEC Motion Picture
Experts Group
(VIPEG). VP9, AOMedia Video 1 (AV1) developed by the Alliance for Open Media
Alliance of
Open Media (A0Media), and Essential Video Coding (EVC) are other video coding
standards for
which the techniques described herein can be applied.
[0042] The techniques described herein can be applied to any of the existing
video codecs (e.g.,
High Efficiency Video Coding (HEVC), Advanced Video Coding (AVC), or other
suitable
existing video codec), and/or can be an efficient coding tool for any video
coding standards being
developed and/or future video coding standards, such as, for example, VVC
and/or other video
coding standard in development or to be developed. For example, examples
described herein can
be performed using video codecs such as VVC, HEVC, AVC, and/or extensions
thereof. However,
the techniques and systems described herein may also be applicable to other
coding standards,
codecs, or formats, such as MPEG, JPEG (or other coding standard for still
images), VP9, AV1,
extensions thereof, or other suitable coding standards already available or
not yet available or
developed. For instance, in some examples, the encoding device 104 and/or the
decoding device
112 may operate according to a proprietary video codec/format, such as AV1,
extensions of AVI,
and/or successor versions of AV1 (e.g., AV2), or other proprietary formats or
industry standards.
Accordingly, while the techniques and systems described herein may be
described with reference
to a particular video coding standard, one of ordinary skill in the art will
appreciate that the
description should not be interpreted to apply only to that particular
standard.
[0043] Referring to FIG. 1, a video source 102 may provide the video data to
the encoding device
104. The video source 102 may be part of the source device, or may be part of
a device other than
the source device. The video source 102 may include a video capture device
(e.g., a video camera,
a camera phone, a video phone, or the like), a video archive containing stored
video, a video server
or content provider providing video data, a video feed interface receiving
video from a video server
or content provider, a computer graphics system for generating computer
graphics video data, a
combination of such sources, or any other suitable video source.
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0044] The video data from the video source 102 may include one or more input
pictures or
frames. A picture or frame is a still image that, in some cases, is part of a
video. In some examples,
data from the video source 102 can be a still image that is not a part of a
video. In HEVC, VVC,
and other video coding specifications, a video sequence can include a series
of pictures. A picture
may include three sample arrays, denoted SL, SCb, and SCr. SL is a two-
dimensional array of
luma samples, SCb is a two-dimensional array of Cb chrominance samples, and
SCr is a two-
dimensional array of Cr chrominance samples. Chrominance samples may also be
referred to
herein as "chroma" samples. A pixel can refer to all three components (luma
and chroma samples)
for a given location in an array of a picture. In other instances, a picture
may be monochrome and
may only include an array of luma samples, in which case the terms pixel and
sample can be used
interchangeably. With respect to example techniques described herein that
refer to individual
samples for illustrative purposes, the same techniques can be applied to
pixels (e.g., all three
sample components for a given location in an array of a picture). With respect
to example
techniques described herein that refer to pixels (e.g., all three sample
components for a given
location in an array of a picture) for illustrative purposes, the same
techniques can be applied to
individual samples.
[0045] The encoder engine 106 (or encoder) of the encoding device 104 encodes
the video data
to generate an encoded video bitstream. In some examples, an encoded video
bitstream (or "video
bitstream" or "bitstream") is a series of one or more coded video sequences. A
coded video
sequence (CVS) includes a series of access units (AUs) starting with an AU
that has a random
access point picture in the base layer and with certain properties up to and
not including a next AU
that has a random access point picture in the base layer and with certain
properties. For example,
the certain properties of a random access point picture that starts a CVS may
include a RASL flag
(e.g., NoRaslOutputFlag) equal to 1. Otherwise, a random access point picture
(with RASL flag
equal to 0) does not start a CVS. An access unit (AU) includes one or more
coded pictures and
control information corresponding to the coded pictures that share the same
output time. Coded
slices of pictures are encapsulated in the bitstream level into data units
called network abstraction
layer (NAL) units. For example, an HEVC video bitstream may include one or
more CVSs
including NAL units. Each of the NAL units has a NAL unit header. In one
example, the header
is one-byte for H.264/AVC (except for multi-layer extensions) and two-byte for
HEVC. The
syntax elements in the NAL unit header take the designated bits and therefore
are visible to all
11
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
kinds of systems and transport layers, such as Transport Stream, Real-time
Transport (RTP)
Protocol, File Format, among others.
[0046] Two classes of NAL units exist in the HEVC standard, including video
coding layer
(VCL) NAL units and non-VCL NAL units. VCL NAL units include coded picture
data forming
a coded video bitstream. For example, a sequence of bits forming the coded
video bitstream is
present in VCL NAL units. A VCL NAL unit can include one slice or slice
segment (described
below) of coded picture data, and a non-VCL NAL unit includes control
information that relates
to one or more coded pictures. In some cases, a NAL unit can be referred to as
a packet. An HEVC
AU includes VCL NAL units containing coded picture data and non-VCL NAL units
(if any)
corresponding to the coded picture data. Non-VCL NAL units may contain
parameter sets with
high-level information relating to the encoded video bitstream, in addition to
other information.
For example, a parameter set may include a video parameter set (VPS), a
sequence parameter set
(SPS), and a picture parameter set (PPS). In some cases, each slice or other
portion of a bitstream
can reference a single active PPS, SPS, and/or VPS to allow the decoding
device 112 to access
information that may be used for decoding the slice or other portion of the
bitstream.
[0047] NAL units may contain a sequence of bits forming a coded representation
of the video
data (e.g., an encoded video bitstream, a CVS of a bitstream, or the like),
such as coded
representations of pictures in a video. The encoder engine 106 generates coded
representations of
pictures by partitioning each picture into multiple slices. A slice is
independent of other slices so
that information in the slice is coded without dependency on data from other
slices within the same
picture. A slice includes one or more slice segments including an independent
slice segment and,
if present, one or more dependent slice segments that depend on previous slice
segments.
[0048] In HEVC, the slices are then partitioned into coding tree blocks (CTBs)
of luma samples
and chroma samples. A CTB of luma samples and one or more CTBs of chroma
samples, along
with syntax for the samples, are referred to as a coding tree unit (CTU). A
CTU may also be
referred to as a "tree block" or a "largest coding unit" (LCU). A CTU is the
basic processing unit
for HEVC encoding. A CTU can be split into multiple coding units (CUs) of
varying sizes. A CU
contains luma and chroma sample arrays that are referred to as coding blocks
(CBs).
12
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0049] The luma and chroma CBs can be further split into prediction blocks
(PBs). A PB is a
block of samples of the luma component or a chroma component that uses the
same motion
parameters for inter-prediction or intra-block copy (IBC) prediction (when
available or enabled
for use). The luma PB and one or more chroma PBs, together with associated
syntax, form a
prediction unit (PU). For inter-prediction, a set of motion parameters (e.g.,
one or more motion
vectors, reference indices, or the like) is signaled in the bitstream for each
PU and is used for inter-
prediction of the luma PB and the one or more chroma PBs. The motion
parameters can also be
referred to as motion information. A CB can also be partitioned into one or
more transform blocks
(TBs). A TB represents a square block of samples of a color component on which
a residual
transform (e.g., the same two-dimensional transform in some cases) is applied
for coding a
prediction residual signal. A transform unit (TU) represents the TBs of luma
and chroma samples,
and corresponding syntax elements. Transform coding is described in more
detail below.
[0050] A size of a CU corresponds to a size of the coding mode and may be
square in shape. For
example, a size of a CU may be 8 x 8 samples, 16 x 16 samples, 32 x 32
samples, 64 x 64 samples,
or any other appropriate size up to the size of the corresponding CTU. The
phrase "N x N" is used
herein to refer to pixel dimensions of a video block in terms of vertical and
horizontal dimensions
(e.g., 8 pixels x 8 pixels). The pixels in a block may be arranged in rows and
columns. In some
implementations, blocks may not have the same number of pixels in a horizontal
direction as in a
vertical direction. Syntax data associated with a CU may describe, for
example, partitioning of the
CU into one or more PUs. Partitioning modes may differ between whether the CU
is intra-
prediction mode encoded or inter-prediction mode encoded. PUs may be
partitioned to be non-
square in shape. Syntax data associated with a CU may also describe, for
example, partitioning of
the CU into one or more TUs according to a CTU. A TU can be square or non-
square in shape.
[0051] According to the HEVC standard, transformations may be performed using
transform
units (TUs). TUs may vary for different CUs. The TUs may be sized based on the
size of PUs
within a given CU. The TUs may be the same size or smaller than the PUs. In
some examples,
residual samples corresponding to a CU may be subdivided into smaller units
using a quadtree
structure known as residual quad tree (RQT). Leaf nodes of the RQT may
correspond to TUs. Pixel
difference values associated with the TUs may be transformed to produce
transform coefficients.
The transform coefficients may then be quantized by the encoder engine 106.
13
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0052] Once the pictures of the video data are partitioned into CUs, the
encoder engine 106
predicts each PU using a prediction mode. The prediction unit or prediction
block is then subtracted
from the original video data to get residuals (described below). For each CU,
a prediction mode
may be signaled inside the bitstream using syntax data. A prediction mode may
include intra-
predi cti on (or intra-picture prediction) or inter-prediction (or inter-
picture prediction). Intra-
prediction utilizes the correlation between spatially neighboring samples
within a picture. For
example, using intra-prediction, each PU is predicted from neighboring image
data in the same
picture using, for example, DC prediction to find an average value for the PU,
planar prediction to
fit a planar surface to the PU, direction prediction to extrapolate from
neighboring data, or any
other suitable types of prediction. Inter-prediction uses the temporal
correlation between pictures
in order to derive a motion-compensated prediction for a block of image
samples. For example,
using inter-prediction, each PU is predicted using motion compensation
prediction from image
data in one or more reference pictures (before or after the current picture in
output order). The
decision whether to code a picture area using inter-picture or intra-picture
prediction may be made,
for example, at the CU level.
[0053] The encoder engine 106 and decoder engine 116 (described in more detail
below) may
be configured to operate according to VVC. According to VVC, a video coder
(such as encoder
engine 106 and/or decoder engine 116) partitions a picture into a plurality of
coding tree units
(CTUs) (where a CTB of luma samples and one or more CTBs of chroma samples,
along with
syntax for the samples, are referred to as a CTU). The video coder can
partition a CTU according
to a tree structure, such as a quadtree-binary tree (QTBT) structure or Multi-
Type Tree (MTT)
structure. The QTBT structure removes the concepts of multiple partition
types, such as the
separation between CUs, PUs, and TUs of HEVC. A QTBT structure includes two
levels,
including a first level partitioned according to quadtree partitioning, and a
second level partitioned
according to binary tree partitioning. A root node of the QTBT structure
corresponds to a CTU.
Leaf nodes of the binary trees correspond to coding units (CUs).
[0054] In an MTT partitioning structure, blocks may be partitioned using a
quadtree partition, a
binary tree partition, and one or more types of triple tree partitions. A
triple tree partition is a
partition where a block is split into three sub-blocks. In some examples, a
triple tree partition
divides a block into three sub-blocks without dividing the original block
through the center. The
14
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
partitioning types in MTT (e.g., quadtree, binary tree, and tripe tree) may be
symmetrical or
asymmetrical.
[0055] When operating according to the AV1 codec, encoding device 104 and
decoding device
112 may be configured to code video data in blocks. In AV1, the largest coding
block that can be
processed is called a superblock. In AV1, a superblock can be either 128x128
luma samples or
64x64 luma samples. However, in successor video coding formats (e.g., AV2), a
superblock may
be defined by different (e.g., larger) luma sample sizes. In some examples, a
superblock is the top
level of a block quadtree. Encoding device 104 may further partition a
superblock into smaller
coding blocks. Encoding device 104 may partition a superblock and other coding
blocks into
smaller blocks using square or non-square partitioning. Non-square blocks may
include N/2xN,
NxN/2, N/4xN, and NxN/4 blocks. Encoding device 104 and decoding device 112
may perform
separate prediction and transform processes on each of the coding blocks.
[0056] AV1 also defines a tile of video data. A tile is a rectangular array of
superblocks that may
be coded independently of other tiles. That is, encoding device 104 and
decoding device 112 may
encode and decode, respectively, coding blocks within a tile without using
video data from other
tiles. However, encoding device 104 and decoding device 112 may perform
filtering across tile
boundaries. Tiles may be uniform or non-uniform in size. Tile-based coding may
enables parallel
processing and/or multi-threading for encoder and decoder implementations.
[0057] In some examples, the video coder can use a single QTBT or MTT
structure to represent
each of the luminance and chrominance components, while in other examples, the
video coder can
use two or more QTBT or MTT structures, such as one QTBT or MTT structure for
the luminance
component and another QTBT or MTT structure for both chrominance components
(or two QTBT
and/or MTT structures for respective chrominance components).
[0058] The video coder can be configured to use quadtree partitioning, QTBT
partitioning, MTT
partitioning, superblock partitioning, or other partitioning structure.
[0059] In some examples, the one or more slices of a picture are assigned a
slice type. Slice
types include an intra-coded slice (I-slice), an inter-coded P-slice, and an
inter-coded B-slice. An
I-slice (intra-coded frames, independently decodable) is a slice of a picture
that is only coded by
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
intra-prediction, and therefore is independently decodable since the I-slice
requires only the data
within the frame to predict any prediction unit or prediction block of the
slice. A P-slice (uni-
directional predicted frames) is a slice of a picture that may be coded with
intra-prediction and
with uni-directional inter-prediction. Each prediction unit or prediction
block within a P-slice is
either coded with intra-prediction or inter-prediction. When the inter-
prediction applies, the
prediction unit or prediction block is only predicted by one reference
picture, and therefore
reference samples are only from one reference region of one frame. A B-slice
(bi-directional
predictive frames) is a slice of a picture that may be coded with intra-
prediction and with inter-
prediction (e.g., either bi-prediction or uni-prediction). A prediction unit
or prediction block of a
B-slice may be bi-directionally predicted from two reference pictures, where
each picture
contributes one reference region and sample sets of the two reference regions
are weighted (e.g.,
with equal weights or with different weights) to produce the prediction signal
of the bi-directional
predicted block. As explained above, slices of one picture are independently
coded. In some cases,
a picture can be coded as just one slice.
[0060] As noted above, intra-picture prediction of a picture utilizes the
correlation between
spatially neighboring samples within the picture. There is a plurality of
intra-prediction modes
(also referred to as "intra modes"). In some examples, the intra prediction of
a luma block includes
35 modes, including the Planar mode, DC mode, and 33 angular modes (e.g.,
diagonal intra
prediction modes and angular modes adjacent to the diagonal intra prediction
modes). The 35
modes of the intra prediction are indexed as shown in Table 1 below. In other
examples, more intra
modes may be defined including prediction angles that may not already be
represented by the 33
angular modes. In other examples, the prediction angles associated with the
angular modes may
be different from those used in REVC.
Intra-prediction mode Associated name
0 INTRA PLANAR
1 INTRA DC
2..34 INTRA ANGULAR2..INTRA ANGULAR34
Table 1 ¨ Specification of intra prediction mode and associated names
16
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0061] Inter-picture prediction uses the temporal correlation between pictures
in order to derive
a motion-compensated prediction for a block of image samples. Using a
translational motion
model, the position of a block in a previously decoded picture (a reference
picture) is indicated by
a motion vector (Ax, Ay), with Ax specifying the horizontal displacement and
Ay specifying the
vertical displacement of the reference block relative to the position of the
current block. In some
cases, a motion vector (Ax, Ay) can be in integer sample accuracy (also
referred to as integer
accuracy), in which case the motion vector points to the integer-pel grid (or
integer-pixel sampling
grid) of the reference frame. In some cases, a motion vector (Ax, Ay) can be
of fractional sample
accuracy (also referred to as fractional-pel accuracy or non-integer accuracy)
to more accurately
capture the movement of the underlying object, without being restricted to the
integer-pel grid of
the reference frame. Accuracy of motion vectors may be expressed by the
quantization level of the
motion vectors. For example, the quantization level may be integer accuracy
(e.g., 1-pixel) or
fractional-pel accuracy (e.g., 1/4-pixel, 1/2-pixel, or other sub-pixel
value). Interpolation is applied
on reference pictures to derive the prediction signal when the corresponding
motion vector has
fractional sample accuracy. For example, samples available at integer
positions can be filtered
(e.g., using one or more interpolation filters) to estimate values at
fractional positions. The
previously decoded reference picture is indicated by a reference index
(refldx) to a reference
picture list. The motion vectors and reference indices can be referred to as
motion parameters. Two
kinds of inter-picture prediction can be performed, including uni-prediction
and bi-prediction.
[0062] With inter-prediction using bi-prediction (also referred to as bi-
directional inter-
prediction), two sets of motion parameters (Axo, yo,refIclx0 and Axi,
yi,refIclxi) are used to
generate two motion compensated predictions (from the same reference picture
or possibly from
different reference pictures). For example, with bi-prediction, each
prediction block uses two
motion compensated prediction signals, and generates B prediction units. The
two motion
compensated predictions are then combined to get the final motion compensated
prediction. For
example, the two motion compensated predictions can be combined by averaging.
In another
example, weighted prediction can be used, in which case different weights can
be applied to each
motion compensated prediction. The reference pictures that can be used in bi-
prediction are stored
in two separate lists, denoted as list 0 and list 1. Motion parameters can be
derived at the encoder
using a motion estimation process.
17
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0063] With inter-prediction using uni-prediction (also referred to as uni-
directional inter-
prediction), one set of motion parameters (Axo, yo,refIclx0) is used to
generate a motion
compensated prediction from a reference picture. For example, with uni-
prediction, each
prediction block uses at most one motion compensated prediction signal, and
generates P
prediction units.
[0064] A PU may include the data (e.g., motion parameters or other suitable
data) related to the
prediction process. For example, when the PU is encoded using intra-
prediction, the PU may
include data describing an intra-prediction mode for the PU. As another
example, when the PU is
encoded using inter-prediction, the PU may include data defining a motion
vector for the PU. The
data defining the motion vector for a PU may describe, for example, a
horizontal component of
the motion vector (ax), a vertical component of the motion vector (Ay), a
resolution for the motion
vector (e.g., integer precision, one-quarter pixel precision or one-eighth
pixel precision), a
reference picture to which the motion vector points, a reference index, a
reference picture list (e.g.,
List 0, List 1, or List C) for the motion vector, or any combination thereof.
[0065] AV1 includes two general techniques for encoding and decoding a coding
block of video
data. The two general techniques are intra prediction (e.g., intra frame
prediction or spatial
prediction) and inter prediction (e.g., inter frame prediction or temporal
prediction). In the context
of AV1, when predicting blocks of a current frame of video data using an intra
prediction mode,
encoding device 104 and decoding device 112 do not use video data from other
frames of video
data. For most intra prediction modes, the video encoding device 104 encodes
blocks of a current
frame based on the difference between sample values in the current block and
predicted values
generated from reference samples in the same frame. The video encoding device
104 determines
predicted values generated from the reference samples based on the intra
prediction mode.
[0066] After performing prediction using intra- and/or inter-prediction, the
encoding device 104
can perform transformation and quantization. For example, following
prediction, the encoder
engine 106 may calculate residual values corresponding to the PU. Residual
values may comprise
pixel difference values between the current block of pixels being coded (the
PU) and the prediction
block used to predict the current block (e.g., the predicted version of the
current block). For
example, after generating a prediction block (e.g., issuing inter-prediction
or intra-prediction), the
18
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
encoder engine 106 can generate a residual block by subtracting the prediction
block produced by
a prediction unit from the current block. The residual block includes a set of
pixel difference values
that quantify differences between pixel values of the current block and pixel
values of the
prediction block. In some examples, the residual block may be represented in a
two-dimensional
block format (e.g., a two-dimensional matrix or array of pixel values). In
such examples, the
residual block is a two-dimensional representation of the pixel values.
[0067] Any residual data that may be remaining after prediction is performed
is transformed
using a block transform, which may be based on discrete cosine transform,
discrete sine transform,
an integer transform, a wavelet transform, other suitable transform function,
or any combination
thereof. In some cases, one or more block transforms (e.g., sizes 32 x 32, 16
x 16, 8 x 8, 4 x 4, or
other suitable size) may be applied to residual data in each CU. In some
aspects, a TU may be used
for the transform and quantization processes implemented by the encoder engine
106. A given CU
having one or more PUs may also include one or more TUs. As described in
further detail below,
the residual values may be transformed into transform coefficients using the
block transforms, and
then may be quantized and scanned using TUs to produce serialized transform
coefficients for
entropy coding.
[0068] In some aspects following intra-predictive or inter-predictive coding
using PUs of a CU,
the encoder engine 106 may calculate residual data for the TUs of the CU. The
PUs may comprise
pixel data in the spatial domain (or pixel domain). The TUs may comprise
coefficients in the
transform domain following application of a block transform. As previously
noted, the residual
data may correspond to pixel difference values between pixels of the unencoded
picture and
prediction values corresponding to the PUs. Encoder engine 106 may form the
TUs including the
residual data for the CU, and may then transform the TUs to produce transform
coefficients for
the CU.
[0069] The encoder engine 106 may perform quantization of the transform
coefficients.
Quantization provides further compression by quantizing the transform
coefficients to reduce the
amount of data used to represent the coefficients. For example, quantization
may reduce the bit
depth associated with some or all of the coefficients. In one example, a
coefficient with an n-bit
value may be rounded down to an m-bit value during quantization, with n being
greater than m.
19
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0070] Once quantization is performed, the coded video bitstream includes
quantized transform
coefficients, prediction information (e.g., prediction modes, motion vectors,
block vectors, or the
like), partitioning information, and any other suitable data, such as other
syntax data. The different
elements of the coded video bitstream may then be entropy encoded by the
encoder engine 106. In
some examples, the encoder engine 106 may utilize a predefined scan order to
scan the quantized
transform coefficients to produce a serialized vector that can be entropy
encoded. In some
examples, encoder engine 106 may perform an adaptive scan. After scanning the
quantized
transform coefficients to form a vector (e.g., a one-dimensional vector), the
encoder engine 106
may entropy encode the vector. For example, the encoder engine 106 may use
context adaptive
variable length coding, context adaptive binary arithmetic coding, syntax-
based context-adaptive
binary arithmetic coding, probability interval partitioning entropy coding, or
another suitable
entropy encoding technique.
[0071] The output 110 of the encoding device 104 may send the NAL units making
up the
encoded video bitstream data over the communications link 120 to the decoding
device 112 of the
receiving device. The input 114 of the decoding device 112 may receive the NAL
units. The
communications link 120 may include a channel provided by a wireless network,
a wired network,
or a combination of a wired and wireless network. A wireless network may
include any wireless
interface or combination of wireless interfaces and may include any suitable
wireless network
(e.g., the Internet or other wide area network, a packet-based network,
WiFiTm, radio frequency
(RF), ultra-wideband (UWB), WiFi-Direct, cellular, Long-Term Evolution (LTE),
WiMaxTm, or
the like). A wired network may include any wired interface (e.g., fiber,
ethernet, powerline
ethernet, ethernet over coaxial cable, digital signal line (DSL), or the
like). The wired and/or
wireless networks may be implemented using various equipment, such as base
stations, routers,
access points, bridges, gateways, switches, or the like. The encoded video
bitstream data may be
modulated according to a communication standard, such as a wireless
communication protocol,
and transmitted to the receiving device.
[0072] In some examples, the encoding device 104 may store encoded video
bitstream data in
storage 108. The output 110 may retrieve the encoded video bitstream data from
the encoder engine
106 or from the storage 108. Storage 108 may include any of a variety of
distributed or locally
accessed data storage media. For example, the storage 108 may include a hard
drive, a storage
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
disc, flash memory, volatile or non-volatile memory, or any other suitable
digital storage media
for storing encoded video data. The storage 108 can also include a decoded
picture buffer (DPB)
for storing reference pictures for use in inter-prediction. In a further
example, the storage 108 can
correspond to a file server or another intermediate storage device that may
store the encoded video
generated by the source device. In such cases, the receiving device including
the decoding device
112 can access stored video data from the storage device via streaming or
download. The file
server may be any type of server capable of storing encoded video data and
transmitting that
encoded video data to the receiving device. Example file servers include a web
server (e.g., for a
website), an FTP server, network attached storage (NAS) devices, or a local
disk drive. The
receiving device may access the encoded video data through any standard data
connection,
including an Internet connection. This may include a wireless channel (e.g., a
Wi-Fi connection),
a wired connection (e.g., DSL, cable modem, etc.), or a combination of both
that is suitable for
accessing encoded video data stored on a file server. The transmission of
encoded video data from
the storage 108 may be a streaming transmission, a download transmission, or a
combination
thereof.
[0073] The input 114 of the decoding device 112 receives the encoded video
bitstream data and
may provide the video bitstream data to the decoder engine 116, or to storage
118 for later use by
the decoder engine 116. For example, the storage 118 can include a decoded
picture buffer (DPB)
for storing reference pictures for use in inter-prediction. The receiving
device including the
decoding device 112 can receive the encoded video data to be decoded via the
storage 108. The
encoded video data may be modulated according to a communication standard,
such as a wireless
communication protocol, and transmitted to the receiving device. The
communication medium for
transmitting the encoded video data can comprise any wireless or wired
communication medium,
such as a radio frequency (RF) spectrum or one or more physical transmission
lines. The
communication medium may form part of a packet-based network, such as a local
area network, a
wide-area network, or a global network such as the Internet. The communication
medium may
include routers, switches, base stations, or any other equipment that may be
useful to facilitate
communication from the source device to the receiving device.
[0074] The decoder engine 116 may decode the encoded video bitstream data by
entropy
decoding (e.g., using an entropy decoder) and extracting the elements of one
or more coded video
21
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
sequences making up the encoded video data. The decoder engine 116 may then
rescale and
perform an inverse transform on the encoded video bitstream data. Residual
data is then passed to
a prediction stage of the decoder engine 116. The decoder engine 116 then
predicts a block of
pixels (e.g., a PU). In some examples, the prediction is added to the output
of the inverse transform
(the residual data).
[0075] The video decoding device 112 may output the decoded video to a video
destination
device 122, which may include a display or other output device for displaying
the decoded video
data to a consumer of the content. In some aspects, the video destination
device 122 may be part
of the receiving device that includes the decoding device 112. In some
aspects, the video
destination device 122 may be part of a separate device other than the
receiving device.
[0076] In some aspects, the video encoding device 104 and/or the video
decoding device 112
may be integrated with an audio encoding device and audio decoding device,
respectively. The
video encoding device 104 and/or the video decoding device 112 may also
include other hardware
or software that is necessary to implement the coding techniques described
above, such as one or
more microprocessors, digital signal processors (DSPs), application specific
integrated circuits
(ASICs), field programmable gate arrays (FPGAs), discrete logic, software,
hardware, firmware
or any combinations thereof. The video encoding device 104 and the video
decoding device 112
may be integrated as part of a combined encoder/decoder (codec) in a
respective device.
[0077] The example system shown in FIG. 1 is one illustrative example that can
be used herein.
Techniques for processing video data using the techniques described herein can
be performed by
any digital video encoding and/or decoding device. Although generally the
techniques of this
disclosure are performed by a video encoding device or a video decoding
device, the techniques
may also be performed by a combined video encoder-decoder, typically referred
to as a "CODEC."
Moreover, the techniques of this disclosure may also be performed by a video
preprocessor. The
source device and the receiving device are merely examples of such coding
devices in which the
source device generates coded video data for transmission to the receiving
device. In some
examples, the source and receiving devices may operate in a substantially
symmetrical manner
such that each of the devices include video encoding and decoding components.
Hence, example
22
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
systems may support one-way or two-way video transmission between video
devices, e.g., for
video streaming, video playback, video broadcasting, or video telephony.
[0078] This disclosure may generally refer to "signaling" certain information,
such as syntax
elements. The term "signaling" may generally refer to the communication of
values for syntax
elements and/or other data used to decode encoded video data. For example, the
video encoding
device 104 may signal values for syntax elements in the bitstream. In general,
signaling refers to
generating a value in the bitstream. As noted above, video source 102 may
transport the bitstream
to video destination device 122 substantially in real time, or not in real
time, such as might occur
when storing syntax elements to storage 108 for later retrieval by the video
destination device 122.
[0079] A video bitstream can also include Supplemental Enhancement Information
(SET)
messages. For example, an SET NAL unit can be part of the video bitstream. In
some cases, an SET
message can contain information that is not needed by the decoding process.
For example, the
information in an SET message may not be essential for the decoder to decode
the video pictures
of the bitstream, but the decoder can use the information to improve the
display or processing of
the pictures (e.g., the decoded output). The information in an SET message can
be embedded
metadata. In one illustrative example, the information in an SET message could
be used by decoder-
side entities to improve the viewability of the content. In some instances,
certain application
standards may mandate the presence of such SET messages in the bitstream so
that the
improvement in quality can be brought to all devices that conform to the
application standard (e.g.,
the carriage of the frame-packing SET message for frame-compatible plano-
stereoscopic 3DTV
video format, where the SET message is carried for every frame of the video,
handling of a recovery
point SET message, use of pan-scan scan rectangle SET message in DVB, in
addition to many other
examples).
[0080] As described above, for each block, a set of motion information (also
referred to herein
as motion parameters) can be available. A set of motion information contains
motion information
for forward and backward prediction directions. The forward and backward
prediction directions
are two prediction directions of a bi-directional prediction mode, in which
case the terms "forward"
and "backward" do not necessarily have a geometrical meaning. Instead,
"forward" and
"backward" correspond to reference picture list 0 (RefPicListO or LO) and
reference picture list 1
23
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
(RefPicListl or L1) of a current picture. In some examples, when only one
reference picture list is
available for a picture or slice, only RefPicListO is available and the motion
information of each
block of a slice is always forward.
[0081] In some cases, a motion vector together with its reference index is
used in coding
processes (e.g., motion compensation). Such a motion vector with the
associated reference index
is denoted as a uni-predictive set of motion information. For each prediction
direction, the motion
information can contain a reference index and a motion vector. In some cases,
for simplicity, a
motion vector itself may be referred in a way that it is assumed that it has
an associated reference
index. A reference index is used to identify a reference picture in the
current reference picture list
(RefPicListO or RefPicList1). A motion vector has a horizontal and a vertical
component that
provide an offset from the coordinate position in the current picture to the
coordinates in the
reference picture identified by the reference index. For example, a reference
index can indicate a
particular reference picture that should be used for a block in a current
picture, and the motion
vector can indicate where in the reference picture the best-matched block (the
block that best
matches the current block) is in the reference picture.
[0082] In H.264/AVC, each inter macroblock (MB) may be partitioned in four
different ways,
including: one 16x16 MB partition; two 16x8 MB partitions; two 8x16 MB
partitions; and four
8x8 MB partitions. Different MB partitions in one MB may have different
reference index values
for each direction (RefPicListO or RefPicList1). In some cases, when an MB is
not partitioned into
four 8x8 MB partitions, it can have only one motion vector for each MB
partition in each direction.
In some cases, when an MB is partitioned into four 8x8 MB partitions, each 8x8
MB partition can
be further partitioned into sub-blocks, in which case each sub-block can have
a different motion
vector in each direction. In some examples, there are four different ways to
get sub-blocks from
an 8x8 MB partition, including: one 8x8 sub-block; two 8x4 sub-blocks; two 4x8
sub-blocks; and
four 4x4 sub-blocks. Each sub-block can have a different motion vector in each
direction.
Therefore, a motion vector is present in a level equal to higher than sub-
block.
[0083] In AVC, a temporal direct mode can be enabled at either the MB level or
the MB partition
level for skip and/or direct mode in B slices. For each MB partition, the
motion vectors of the
block co-located with the current MB partition in the RefPicListl[ 0 ] of the
current block are used
24
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
to derive the motion vectors. Each motion vector in the co-located block is
scaled based on POC
distances.
[0084] A spatial direct mode can also be performed in AVC. For example, in
AVC, a direct
mode can also predict motion information from the spatial neighbors.
[0085] As noted above, in HEVC, the largest coding unit in a slice is called a
coding tree block
(CTB). A CTB contains a quad-tree, the nodes of which are coding units. The
size of a CTB can
range from 16x16 to 64x64 in the HEVC main profile. In some cases, 8x8 CTB
sizes can be
supported. A coding unit (CU) could be the same size of a CTB and as small as
8x8. In some cases,
each coding unit is coded with one mode. When a CU is inter-coded, the CU may
be further
partitioned into 2 or 4 prediction units (PUs), or may become just one PU when
further partition
does not apply. When two PUs are present in one CU, they can be half size
rectangles or two
rectangles with 1/4 or % size of the CU. When the CU is inter-coded, one set
of motion information
is present for each PU. In addition, each PU is coded with a unique inter-
prediction mode to derive
the set of motion information.
[0086] For motion prediction in HEVC for example, there are two inter-
prediction modes,
including merge mode and advanced motion vector prediction (AMVP) mode for a
prediction unit
(PU). Skip is considered as a special case of merge. In either AMVP or merge
mode, a motion
vector (MV) candidate list is maintained for multiple motion vector
predictors. The motion
vector(s), as well as reference indices in the merge mode, of the current PU
are generated by taking
one candidate from the MV candidate list. In some examples, one or more
scaling window offsets
can be included along with stored motion vectors in a MV candidate list.
[0087] In examples where a MV candidate list is used for motion prediction of
a block, the MV
candidate list may be constructed by the encoding device and the decoding
device separately. For
instance, the MV candidate list can be generated by an encoding device when
encoding a block,
and can be generated by a decoding device when decoding the block. Information
related to motion
information candidates in the MV candidate list (e.g., information related to
one or more motion
vectors, information related to one or more LIC flags which can be stored in
the MV candidate list
in some cases, and/or other information), can be signaled between the encoding
device and the
decoding device. For example, in the merge mode, index values to the stored
motion information
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
candidates can be signaled from an encoding device to a decoding device (e.g.,
in a syntax
structure, such as the picture parameter set (PPS), sequence parameter set
(SPS), video parameter
set (VPS), a slice header, a supplemental enhancement information (SET)
message sent in or
separately from the video bitstream, and/or other signaling). The decoding
device can construct a
MV candidate list and use the signaled references or indexes to obtain one or
more motion
information candidates from the constructed MV candidate list to use for
motion compensation
prediction. For example, the decoding device 112 may construct a MV candidate
list and use a
motion vector (and in some cases an LIC flag) from an indexed location for
motion prediction of
the block. In the case of AMVP mode, in addition to the references or indexes,
differences or
residual values may also be signaled as deltas. For example, for the AMVP
mode, the decoding
device can construct one or more MV candidate lists and apply the delta values
to one or more
motion information candidates obtained using the signaled index values in
performing motion
compensation prediction of the block.
[0088] In some examples, the MV candidate list contains up to five candidates
for the merge
mode and two candidates for the AMVP mode. In other examples, different
numbers of candidates
can be included in a MV candidate list for merge mode and/or AMVP mode. A
merge candidate
may contain a set of motion information. For example, a set of motion
information can include
motion vectors corresponding to both reference picture lists (list 0 and list
1) and the reference
indices. If a merge candidate is identified by a merge index, the reference
pictures are used for the
prediction of the current blocks, as well as the associated motion vectors are
determined. However,
under AMVP mode, for each potential prediction direction from either list 0 or
list 1, a reference
index needs to be explicitly signaled, together with an MVP index to the MV
candidate list since
the AMVP candidate contains only a motion vector. In AMVP mode, the predicted
motion vectors
can be further refined.
[0089] As can be seen above, a merge candidate corresponds to a full set of
motion information,
while an AMVP candidate contains just one motion vector for a specific
prediction direction and
reference index. The candidates for both modes are derived similarly from the
same spatial and
temporal neighboring blocks.
26
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0090] In some examples, merge mode allows an inter-predicted PU to inherit
the same motion
vector or vectors, prediction direction, and reference picture index or
indices from an inter-
predicted PU that includes a motion data position selected from a group of
spatially neighboring
motion data positions and one of two temporally co-located motion data
positions. For AMVP
mode, motion vector or vectors of a PU can be predicatively coded relative to
one or more motion
vector predictors (MVPs) from an AMVP candidate list constructed by an encoder
and/or a
decoder. In some instances, for single direction inter-prediction of a PU, the
encoder and/or
decoder can generate a single AMVP candidate list. In some instances, for bi-
directional prediction
of a PU, the encoder and/or decoder can generate two AMVP candidate lists, one
using motion
data of spatial and temporal neighboring PUs from the forward prediction
direction and one using
motion data of spatial and temporal neighboring PUs from the backward
prediction direction.
[0091] The candidates for both modes can be derived from spatial and/or
temporal neighboring
blocks. For example, FIG. 2A and FIG. 2B include conceptual diagrams
illustrating spatial
neighboring candidates. FIG. 2A illustrates spatial neighboring motion vector
(MV) candidates for
merge mode. FIG. 2B illustrates spatial neighboring motion vector (MV)
candidates for AMVP
mode. Spatial MV candidates are derived from the neighboring blocks for a
specific PU (PUO),
although the methods generating the candidates from the blocks differ for
merge and AMVP
modes.
[0092] In merge mode, the encoder and/or decoder can form a merging candidate
list by
considering merging candidates from various motion data positions. For
example, as shown in
FIG. 2A, up to four spatial MV candidates can be derived with respect to
spatially neighboring
motion data positions shown with numbers 0-4 in FIG. 2A. The MV candidates can
be ordered in
the merging candidate list in the order shown by the numbers 0-4. For example,
the positions and
order can include: left position (0), above position (1), above right position
(2), below left position
(3), and above left position (4).
[0093] In AVMP mode shown in FIG. 2B, the neighboring blocks are divided into
two groups:
left group including the blocks 0 and 1, and above group including the blocks
2, 3, and 4. For each
group, the potential candidate in a neighboring block referring to the same
reference picture as that
indicated by the signaled reference index has the highest priority to be
chosen to form a final
27
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
candidate of the group. It is possible that all neighboring blocks do not
contain a motion vector
pointing to the same reference picture. Therefore, if such a candidate cannot
be found, the first
available candidate will be scaled to form the final candidate, thus the
temporal distance
differences can be compensated.
[0094] FIG. 3A and FIG. 3B include conceptual diagrams illustrating temporal
motion vector
prediction. A temporal motion vector predictor (TMVP) candidate, if enabled
and available, is
added to a MV candidate list after spatial motion vector candidates. The
process of motion vector
derivation for a TMVP candidate is the same for both merge and AMVP modes. In
some instances,
however, the target reference index for the TMVP candidate in the merge mode
can be set to zero
or can be derived from that of the neighboring blocks.
[0095] The primary block location for TMVP candidate derivation is the bottom
right block
outside of the collocated PU, as shown in FIG. 3A as a block "T", to
compensate for the bias to
the above and left blocks used to generate spatial neighboring candidates.
However, if that block
is located outside of the current CTB (or LCU) row or motion information is
not available, the
block is substituted with a center block of the PU. A motion vector for a TMVP
candidate is derived
from the co-located PU of the co-located picture, indicated in the slice
level. Similar to temporal
direct mode in AVC, a motion vector of the TMVP candidate may be subject to
motion vector
scaling, which is performed to compensate for distance differences.
[0096] Other aspects of motion prediction are covered in the HEVC standard
and/or other
standard, format, or codec. For example, several other aspects of merge and
AMVP modes are
covered. One aspect includes motion vector scaling. With respect to motion
vector scaling, it can
be assumed that the value of motion vectors is proportional to the distance of
pictures in the
presentation time. A motion vector associates two pictures ¨ the reference
picture and the picture
containing the motion vector (namely the containing picture). When a motion
vector is utilized to
predict the other motion vector, the distance of the containing picture and
the reference picture is
calculated based on the Picture Order Count (POC) values.
[0097] For a motion vector to be predicted, both its associated containing
picture and reference
picture may be different. Therefore, a new distance (based on POC) is
calculated. And, the motion
vector is scaled based on these two POC distances. For a spatial neighboring
candidate, the
28
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
containing pictures for the two motion vectors are the same, while the
reference pictures are
different. In HEVC, motion vector scaling applies to both TMVP and AMVP for
spatial and
temporal neighboring candidates.
[0098] Another aspect of motion prediction includes artificial motion vector
candidate
generation. For example, if a motion vector candidate list is not complete,
artificial motion vector
candidates are generated and inserted at the end of the list until all
candidates are obtained. In
merge mode, there are two types of artificial MV candidates: combined
candidate derived only for
B-slices; and zero candidates used only for AMVP if the first type does not
provide enough
artificial candidates. For each pair of candidates that are already in the
candidate list and that have
necessary motion information, bi-directional combined motion vector candidates
are derived by a
combination of the motion vector of the first candidate referring to a picture
in the list 0 and the
motion vector of a second candidate referring to a picture in the list 1.
[0099] In some implementations, a pruning process can be performed when adding
or inserting
new candidates into an MV candidate list. For example, in some cases it is
possible for MV
candidates from different blocks to include the same information. In such
cases, storing duplicative
motion information of multiple MV candidates in the MV candidate list can lead
to redundancy
and a decrease in the efficiency of the MV candidate list. In some examples,
the pruning process
can eliminate or minimize redundancies in the MV candidate list. For example,
the pruning process
can include comparing a potential MV candidate to be added to an MV candidate
list against the
MV candidates which are already stored in the MV candidate list. In one
illustrative example, the
horizontal displacement (Ax) and the vertical displacement (Ay) (indicating a
position of a
reference block relative to a position of the current block) of a stored
motion vector can be
compared to the horizontal displacement (Ax) and the vertical displacement
(Ay) of the motion
vector of a potential candidate. If the comparison reveals that the motion
vector of the potential
candidate does not match any of the one or more stored motion vectors, the
potential candidate is
not considered as a candidate to be pruned and can be added to the MV
candidate list. If a match
is found based on this comparison, the potential MV candidate is not added to
the MV candidate
list, avoiding the insertion of an identical candidate. In some cases, to
reduce complexity, only a
limited number of comparisons are performed during the pruning process instead
of comparing
each potential MV candidate with all existing candidates.
29
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0100] In certain coding schemes such as HEVC, Weighted Prediction (WP) is
supported, in
which case a scaling factor (denoted by a), a shift number (denoted by s) and
an offset (denoted
by b) is used in the motion compensation. Suppose the pixel value in position
(x, y) of the reference
picture is p(x, y), then p'(x, y) = ((a*p(x, y) + (1 << (s-1))) >> s) + b
instead of p(x, y) is used as
the prediction value in motion compensation.
[0101] When WP is enabled, for each reference picture of current slice, a flag
is signaled to
indicate whether WP applies for the reference picture or not. If WP applies
for one reference
picture, a set of WP parameters (i.e., a, s and b) is sent to the decoder and
is used for motion
compensation from the reference picture. In some examples, to flexibly turn
on/off WP for luma
and chroma component, WP flag and WP parameters are separately signaled for
luma and chroma
component. In WP, one same set of WP parameters is used for all pixels in one
reference picture.
[0102] FIG. 4A is a diagram illustrating an example of neighbor reconstructed
samples of a
current block 402 and neighbor samples of a reference block 404 used for uni-
directional inter-
prediction. A motion vector MV can be coded for the current block 402, where
the MV can include
a reference index to a reference picture list and/or other motion information
for identifying the
reference block 404. For example, the MV can include a horizontal and a
vertical component that
provides an offset from the coordinate position in the current picture to the
coordinates in the
reference picture identified by the reference index. FIG. 4B is a diagram
illustrating an example
of neighbor reconstructed samples of a current block 422 and neighbor samples
of a first reference
block 424 and a second reference block 426 used for bi-directional inter-
prediction. In this case,
two motion vectors MVO and MV1 can be coded for the current block 422 to
identify the first
reference block 424 and a second reference block 426, respectively.
[0103] Bilateral matching (BM) is a technique that can be used to refine a
pair of two initial
motion vectors (e.g., a first motion vector MVO and a second motion vector
MV1). For example,
BM can be performed by searching around the pair of initial motion vectors MVO
and MV1 to
derive refined motion vectors (e.g., refined motion vectors MVO' and MV1').
The refined motion
vectors MVO' and MV1' can subsequently be used to replace the first motion
vector MVO and the
second motion vector MV1, respectively. The refined motion vectors can be
selected in a search
as the motion vectors identified in the search that minimize a block matching
cost.
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0104] In some examples, the block matching cost can be generated based on the
similarity
between the two motion compensated predictors generated for the two MVs.
Example criteria for
block matching costs include, but are not limited to, sum of absolute
difference (SAD), sum of
absolute transformed difference (SATD), sum of square error (SSE), etc. Block
matching cost
criteria may also include a regularization term that is derived based on the
MV differences between
the current MV pair (e.g., the MV pair being considered for selection as the
refined motion vectors
MVO' and MV1') and the initial MV pair (e.g., MVO and MV1).
[0105] In some examples, one or more constraints can be applied to the MV
difference terms
MVDO and MVD1 (e.g., where MVDO =MVO' ¨MVO and MVD1 = MV1' ¨MV1). For example,
in some cases, a constraint can be applied based on the assumption that MVDO
and MVD1 are
proportional to the temporal distances (TD) between the current picture (e.g.,
current block) and
the reference pictures (e.g., reference blocks) pointed to by the two MVs. In
some examples, a
constraint can be applied based on the assumption that MVDO = ¨MVD1 (e.g.,
MVDO and MVD1
have equal magnitudes but opposite signs).
[0106] In some examples, inter-predicted CUs can be associated with one or
more motion
parameters. For example, in the Versatile Video Coding standard (VVC), each
inter-predicted CU
can be associated with one or more motion parameters that can include, but are
not limited to,
motion vectors, reference picture indices and reference picture list usage
index. The motion
parameters can further include additional information associated with coding
features of VVC to
be used for inter-predicted sample generation. The motion parameters can be
signaled in an explicit
or implicit manner. For example, when a CU is coded with skip mode, the CU is
associated with
one PU and has no significant residual coefficients and no coded motion vector
delta or reference
picture index.
[0107] In some aspects, a merge mode can be specified wherein the motion
parameters for the
current CU are obtained from neighboring CUs, including spatial and temporal
candidates. One or
more merge modes can additionally, or alternatively, be specified based on
additional schedules
introduced in the VVC standard. In some examples, the merge mode can be
applied to any inter-
predicted CU (e.g., the merge mode can be applied beyond skip mode). In some
examples, an
alternative to merge mode can include the explicit transmission of one or more
motion parameters.
31
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
For example, motion vectors, corresponding reference picture indices for each
reference picture
list, reference picture list usage flags, and other related information can be
signaled explicitly for
each CU.
[0108] Beyond the inter coding features in HEVC, VVC includes a number of new
and refined
inter prediction coding techniques, which include: Extended merge prediction;
Merge mode with
motion vector difference (MMVD); Symmetric MVD (SMVD) signaling; Affine motion
compensated prediction; Sub-block based temporal motion vector prediction
(SbTMVP);
Adaptive motion vector resolution (AMVR); Motion field storage: 1/16th luma
sample MV
storage and 8x8 motion field compression; Bi-prediction with CU-level weight
(BCW); Bi-
directional optical flow (BDOF); Decoder side motion vector refinement (DMVR);
Geometric
partitioning mode (GPM); Combined inter and intra prediction (CIIP).
[0109] For extended merge prediction in VVC merge mode (e.g., referred to as a
regular or
default merge mode), the merge candidate list can be constructed by including
the following five
types of candidates in order: Spatial motion vector prediction (MVP) from
spatial neighbor CUs;
Temporal MVP from collocated CUs; History-based MVP from a first-in-first-out
(FIFO) table;
Pairwise average MVP; and Zero MVs. The size of the merge candidate list can
be signaled in a
sequence parameter set header. A maximum allowed size of the merge candidate
list can be six
(e.g., six entries or six candidates). For each CU coded in merge mode, an
index of the best merge
candidate(s) is encoded using truncated unary binarization (TU). In some
examples, VVC can also
support parallel derivation of the merge candidate lists for all CUs within a
certain size or area
(e.g., as done in HEVC). The five aforementioned types of merge candidates,
and an associated
example derivation process of each category of merge candidate, are described
in turn below.
[0110] FIG. 5 illustrates locations or positions of spatial merge candidates
for use in processing
a block in accordance with some examples of the disclosure. For example, FIG.
5 depicts example
positions of spatial merge candidates (also referred to as "spatial
neighbors") AO, Al, BO, Bl, and
B2 for use in processing block 500 in accordance with some examples of the
disclosure. The spatial
neighbors AO, Al, BO, Bl, and B2 are indicated in FIG. 5 based on their
relationship with block
500. The derivation of spatial merge candidates in VVC can be the same as in
HEVC, with the
positions of the first two merge candidates swapped. In some examples, a
maximum of four merge
32
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
candidates may be selected from the five spatial merge candidates located in
the positions depicted
in FIG. 5 (e.g., AO, Al, BO, Bl, and B2).
[0111] The order of derivation can be BO, AO, Bl, Al and B2. For example, the
merge candidate
position B2 may be considered only when one or more CUs associated with the
positions BO, AO,
Bl, Al are not available or are intra coded. A CU associated with the
positions BO, AO, Bl, or Al
may be unavailable because the CU belongs to a different slice or tile. In
some aspects, after a
merge candidate at position Al is added, the addition of the remaining merge
candidates can be
subject to a redundancy check. The redundancy check can be performed such that
merge
candidates with the same motion information are excluded from the merge
candidate list (e.g., to
improve coding efficiency).
[0112] FIG. 6 illustrates aspects of motion vector scaling 600 for temporal
merge candidates for
use in processing a block in accordance with some examples of the disclosure.
In some examples,
temporal merge candidate derivation can be performed wherein one merge
candidate (e.g., a
temporal merge candidate) is added to the merge candidate list. Temporal merge
candidate
derivation can be performed based on a scaled motion vector. The scaled motion
vector can be
derived based on a collocated CU included in a collocated reference picture.
The reference picture
list to be used for derivation of the collocated CU can be explicitly signaled
in the slice header.
[0113] For example, FIG. 6 depicts a current picture 610 and a collocated
picture 630, which
may be associated with a current reference picture 615 and a collocated
reference picture 635,
respectively. FIG. 6 also depicts a current CU 612 (e.g., associated with
current picture 610) and
a collocated CU (e.g., associated with collocated picture 630). In some
examples, a scaled motion
vector for temporal merge candidate derivation can be derived or obtained as
illustrated in FIG. 6.
For example, FIG. 6 depicts a dotted line 611, which is scaled from the motion
vector of collocated
CU 632 using the Picture Order Count (POC) distances, tb and td. In some
examples, tb is the POC
difference between current reference picture 615 and current picture 610 and
td is the POC
difference between the collocated reference picture 635 and collocated picture
630. The reference
picture index of the temporal merge candidate can be set equal to zero.
[0114] FIG. 7 illustrates aspects of temporal merge candidates 700 for use in
processing a block
in accordance with some examples of the disclosure. In some examples, after
the single temporal
33
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
merge candidate is selected as discussed above with respect to FIG. 6, a
position for the temporal
merge candidate can be selected between candidate positions Co and Ci, as
depicted in FIG. 7. In
some examples, if a CU at position Co is not available, is intra coded, or is
outside of the current
row of CTUs, the candidate position Ci may be used. Otherwise, position Co is
used in the
derivation of the temporal merge candidate.
[0115] In some aspects, history-based motion vector prediction (HMVP) merge
candidates may
be added to the merge candidate list after the spatial MVP merge candidates
(e.g., as described
above with respect to FIG. 5) and the TMVP merge candidates (e.g., as
described above with
respect to FIGS. 6 and 7). HMVP merge candidates can be derived based on the
motion
information of a previously coded block. For example, motion information of a
previously coded
block can be stored (e.g., in a table) and used as a motion vector prediction
(MVP) for the current
CU. In some examples, a table with multiple HMVP candidates can be maintained
during an
encoding and/or decoding process. The table is reset (e.g., emptied) when a
new CTU row is
encountered. Whenever there is a non-sub-block inter-coded CU, the associated
motion
information is added to the last entry of the table as a new HMVP candidate.
[0116] In some examples, an HMVP table size S can be set to a value of six
(e.g., up to six
History-based MVP (HMVP) candidates may be added to the HMVP table). When
inserting a new
HMVP candidate into the HMVP table, a constrained first-in-first-out (FIFO)
rule can be utilized.
The constrained FIFO rule can include a redundancy check applied to determine
whether there is
an identical HMVP in the table (e.g., to determine whether the newly inserted
HMVP candidate is
the same as an existing HMVP candidate in the table). If the redundancy check
for the newly
inserted HMVP candidate finds an identical HMVP is already included in the
table, the identical
HMVP can be removed from the table and all the HMVP candidates afterwards are
moved forward.
[0117] In some aspects, HMVP candidates (e.g., included in the HMVP list or
HMVP table) can
subsequently be used to construct or otherwise generate the merge candidate
list. For example, to
generate the merge candidate list, the latest several HMVP candidates in the
table can be checked
in order and inserted to the merge candidate list after the TMVP candidate. A
redundancy check
can be applied on the HMVP candidates added to the merge candidate list,
wherein the redundancy
check is used to determine whether the HMVP candidates are the same as or
identical to any of
34
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
the spatial or temporal merge candidates that were previously added or already
included in the
merge candidate list.
[0118] In some examples, the number of redundancy check operations performed
in association
with generating the merge candidate list and/or the HMVP table can be reduced.
For example, a
number of HMVP candidates used for merge list generation can be set as (N <= 4
) ? M: (8 ¨ N),
wherein N indicates the number of existing candidates in the merge candidate
list and M indicates
the number of available HMVP candidates in the HMVP table. In other words, if
the condition N
<= 4 evaluates as true (e.g., the merge candidate list contains four
candidates or fewer), all M of
the HMVP candidates in the HMVP table will be used for merge list generation.
If the condition
N <= 4 evaluates as false (e.g., the merge candidate list contains more than
four candidates), then
8 ¨ N of the HMVP candidates in the HMVP table will be used for merge list
generation. In some
examples, once the total number of available merge candidates reaches the
maximally allowed
number of merge candidates minus 1, the merge candidate list construction
process from HMVP
can be terminated.
[0119] Pairwise average merge candidate derivation can be performed based on
pre-defined
pairs of merge candidates in the existing merge candidate list. For example,
pairwise average
merge candidates can be generated by averaging pre-defined pairs of merge
candidates in the
existing merge candidate list, where the pre-defined pairs are given as 1(0,
1), (0, 2), (1, 2), (0, 3),
(1, 3), (2, 3)1, where the numbers denote the merge indices to the merge
candidate list. In some
cases, averaged motion vectors can be calculated separately for each reference
list. If both motion
vectors (e.g., of a pre-defined pair) are included in the same list, these two
motion vectors can be
averaged even when the two motion vectors point to different reference
pictures. In some cases, if
only one motion vector (e.g., of a pre-defined pair) is available, the one
available motion vector
can be used as the averaged motion vector. If no motion vector (e.g., of a pre-
defined pair) is
available, the list can be identified as invalid. In some examples, if the
merge candidate list is not
full after pair-wise average merge candidates are added as described above,
one or more zero
MVPs can be inserted at the end of the merge candidate list until the maximum
number of merge
candidates is reached.
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0120] In some aspects, bi-prediction with CU-level weight (BCW) can be
utilized. For example,
a bi-prediction signal can be generated by averaging two prediction signals
obtained from two
different reference pictures. A bi-prediction signal can additionally, or
alternatively, be generated
using two different motion vectors. In some examples, a bi-prediction signal
can be generated by
averaging two prediction signals obtained from two different reference signals
and/or using two
different motion vectors, using the HEVC standard.
[0121] In other aspects, the bi-prediction mode can be extended beyond simple
averaging to
include weighted averaging of the two prediction signals. For example, the bi-
prediction mode can
include weighted averaging of the two prediction signals using the VVC
standard. In some
examples, the bi-prediction mode can include weighted averaging of the two
prediction signals as
follows:
Pbt-pred = ((8 w) * Po w * P1 + 4) >> 3
Eq. (1)
[0122] The weighted averaging bi-prediction given in Eq. (1) can include five
weights, w c {-
2,3,4,5,10}. For each bi-predicted CU, the weight w can be determined
according to one or more
of the following. In one example, for a non-merge CU, a weight index can be
signaled after the
motion vector difference (MVD). In another example, for a merge CU, the weight
index can be
inferred from neighboring blocks, based on the merge candidate index. In some
cases, BCW may
only be applied to CUs with 256 or more luma samples (e.g., CUs for which the
product of CU
width and CU height is greater than or equal to 256). In some cases, all five
of the weights w may
be used for low-delay pictures. For non-low-delay pictures, only three of the
five weights w may
be used (e.g., the three weights w E {3,4,5}).
[0123] At the encoder, fast search algorithms can be applied to find the
weight index without
significantly increasing the encoder complexity, as will be described in
greater depth below. In
some examples, fast search algorithms can be applied to find the weight index
as described at least
in part in JVET-L0646. For example, when combined with adaptive motion vector
resolution
(AMVR), unequal weights are only conditionally checked for 1-pel and 4-pel
motion vector
precisions if the current picture is a low-delay picture. When combined with
affine mode, affine
motion estimation (ME) can be performed for unequal weights if and only if the
affine mode is
selected as the current best mode. When the two reference pictures in a bi-
prediction mode are the
36
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
same, unequal weights may be only conditionally checked. In some cases,
unequal weights are not
searched when certain conditions are met (e.g., depending on the picture order
count (POC)
distance between the current picture and its reference pictures, the coding
quantization parameter
(QP), and/or the temporal level, etc.)
[0124] In some examples, a BCW weight index can be coded using one context-
coded bin
followed by one or more bypass-coded bins. For example, the first context-
coded bin can be used
to indicate if equal weight is used. If unequal weight is used, additional
bins can be signaled using
bypass coding to indicate which unequal weight is used.
[0125] Weighted prediction (WP) is a coding tool supported by the H.264/AVC
and HEVC
standards to efficiently code video content with fading. Support for WP was
also added into the
VVC standard. WP can be used to allow one or more weighting parameters (e.g.,
weight and offset)
to be signaled for each reference picture in each of the reference picture
lists LO and Li.
Subsequently, during motion compensation, the weight(s) and offset(s) of the
corresponding
reference picture(s) are applied.
[0126] WP and BCW can be utilized with different types of video content. In
some examples, if
a CU utilizes WP, then the BCW weight index may not be signaled and w can be
inferred to have
a value of 4 (e.g., equal weight is applied). For example, the BCW weight
index may not be
signaled when a CU utilizes WP in order to avoid interactions between WP and
BCW (e.g., which
can complicate VVC decoder design).
[0127] For a merge CU, the weight index can be inferred from neighboring
blocks based on the
merge candidate index, in both normal merge mode and inherited affine merge
mode. For
constructed affine merge mode, the affine motion information can be
constructed based on the
motion information of up to three blocks. For example, the BCW index for a CU
using the
constructed affine merge mode can be set equal to the BCW index of the first
control point MV.
In some examples, using the VVC standard, combined inter and intra prediction
(CIIP) and bi-
prediction with CU-level weight (BCW) cannot be jointly applied for a CU. When
a CU is coded
with CIIP mode, the BCW index of the current CU can be set to a value of 2
(e.g., set to an equal
weight).
37
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0128] FIG. 8 is a diagram 800 illustrating aspects of bilateral matching in
accordance with some
examples of the disclosure. As mentioned previously, bilateral matching (BM)
can be used to
refine a pair of two initial motion vectors MVO and MV1. For example, BM can
be performed by
searching around the MVO and MV1 to derive refined motion vectors MVO' and
MV1',
respectively, that minimize a block matching cost. The block matching cost can
be calculated based
on a similarity between two motion compensated predictors generated using the
pair of initial
motion vectors (e.g., MVO and MV1). For example, the block matching cost can
be based on a
sum of absolute differences (SAD). The block matching cost can additionally,
or alternatively, be
based on or include a regularization term that is based on the motion vector
differences (MVDs)
between a current MV pair (e.g., a currently tested MVO' and MV1') and the
initial MV pair (e.g.,
MVO and MV1). As will be described in greater depth below, one or more
constraints can be
applied based on an MVDO (e.g., MVDO = MVO' ¨ MVO) and an MVD1 (e.g., MVD1 =
MV1' ¨
MV1).
[0129] As mentioned previously, in the Versatile Video Coding standard (VVC),
bilateral
matching-based decoder side motion vector refinement (DMVR) can be applied to
increase the
accuracy of (e.g., refine) the MVs of a bi-prediction merge candidate. For
example, as illustrated
in the example of FIG. 8, bilateral matching-based DMVR can be applied to
increase the accuracy
of or otherwise refine the MVs of a bi-prediction merge candidate 812. Bi-
prediction merge
candidate 812 can be included in a current picture 810 and can be associated
with an initial pair of
motion vectors MVO and MV1. The initial motion vectors MVO and MV1 can be
obtained,
identified, or otherwise determined for bi-prediction merge candidate 812
prior to performing
bilateral matching-based DMVR. Subsequently, the initial motion vectors MVO
and MV1 can be
used to identify refined motion vectors MVO' and MV1' for bi-prediction merge
candidate 812, as
will be described in greater depth below.
[0130] As illustrated in the example of FIG. 8, the first initial motion
vector MVO can point to a
first reference picture 830. First reference picture 830 can be associated
with a backward direction
(e.g., relative to current picture 810) and/or can be included in reference
picture list LO. The second
initial motion vector MV1 can point to a second reference picture 820. Second
reference picture
820 can be associated with a forward direction (e.g., relative to current
picture 810) and/or can be
included in reference picture list Li.
38
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0131] The first initial motion vector MVO can be used to determine or
generate a first predictor
832, which can be a block included in first reference picture 830. First
predictor 832 can also be
referred to as a first candidate block (e.g., first predictor 832 is a
candidate block in first reference
picture 830 and/or reference picture list LO). The second initial motion
vector MV1 can be used to
determine or generate a second predictor 822, which can be a block included in
second reference
picture 820. Second predictor 822 can also be referred to as a second
candidate block (e.g., second
predictor 822 is a candidate block in second reference picture 820 and/or
reference picture list L1).
[0132] Subsequently, a search can be performed around the first predictor 832
and the second
predictor 822 to identify or determine the first refined motion vector MVO'
and the second refined
motion vector MV1', respectively. As illustrated, a surrounding area
associated with each of the
first predictor 832 (e.g., an area within first reference picture 830 and/or
reference picture list LO)
and the second predictor 822 (e.g., an area within second reference picture
820 and/or reference
picture list L1) can be searched. For example, the surrounding area associated
with first predictor
832 can be searched to identify or examine one or more refined candidate
blocks 834, and the
surrounding area associated with second predictor 822 can be searched to
identify or examine one
or more refined candidate blocks 824. The search can be performed based on one
or more of a
distortion (e.g., SAD, SATD, SSE, etc.) and/or a regularization term
calculated between one of the
initial predictors 832 or 822, and a corresponding refined candidate block 834
or 824. In some
examples, the distortion and/or regularization can be calculated based on a
distance moved away
from an initial point (e.g., the distance between an initial point associated
with the initial predictor
832 or 822, and a searched point associated with the refined candidate block
834 or 824,
respectively).
[0133] As the search moves around the initial point associated with each of
the initial predictors
832 and 822, a new refined candidate block (e.g., refined candidate blocks 834
and 824) is
obtained. Each new refined candidate block can be associated with a new cost
(e.g., a calculated
SAD value, one of the motion vector differences MVDO or MVD1, etc.). The
search can be
associated with a search range and/or search window. After searching each
candidate block
included in the search range and/or search window for initial predictors 832
and 822 (e.g., and
determining the corresponding cost for each searched candidate block), a
candidate block with the
39
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
minimum determined cost can be identified and used to generate the refined
motion vectors MVO'
and MV1' .
[0134] In some examples, bilateral matching (BM) based DMVR can be performed
by
calculating the SAD between two candidate blocks in the reference picture list
LO and list Ll. As
illustrated in FIG. 8, the SAD between the blocks based on each MV' candidate
(e.g., the blocks
834 and 824) around the initial MV (e.g., around predictors 832 and 822,
respectively) can be
calculated. The MV' candidate with the lowest SAD can be selected as the
refined MV and used
to generate the bi-predicted signal. In some examples, the SAD of the initial
MVs is subtracted by
1/4 of the SAD value to serve as regularization term. In some cases, the
temporal distances (e.g.,
Picture Order Count (POC) difference) from two reference pictures to the
current picture can be
the same, and MVDO and MVD1 can have the same magnitude but opposite sign
(e.g., MVDO = -
MVD1).
[0135] In some cases, bilateral matching-based DMVR can be performed using a
refinement
search range of two integer luma samples from the initial MV. For example, in
the context of FIG.
8, bilateral matching-based DMVR can be performed using a refinement search
range of two
integer luma samples from the initial motion vectors MVO and MV1 (e.g., from
the initial
predictors 832 and 822, respectively). The searching can include an integer
sample offset search
stage and a fractional sample refinement stage.
[0136] In some cases, a 25-point full search can be applied for integer sample
offset searching.
The 25-point full refinement search can be performed by first calculating the
SAD of the initial
MV pair (e.g., the initial MV pair of MVO and MV1 and/or the initial
predictors 832 and 822). If
the SAD of the initial MV pair is smaller than a threshold, the integer sample
stage of DMVR can
be terminated. If the SAD of the initial MV pair is not smaller than the
threshold, SADs of the
remaining 24 points can be calculated and checked in raster scanning order.
The point with the
smallest SAD can then be selected as the output of the integer sample offset
searching stage.
[0137] The integer sample offset search can be followed by fractional sample
refinement, as
mentioned above. In some cases, calculational complexity can be reduced by
using one or more
parametric error surface equations to derive the fractional sample refinement
(e.g., rather than
performing additional search(es) with SAD comparison). The fractional sample
refinement can be
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
conditionally invoked based on the output of the integer sample offset search
stage. For example,
when the integer sample offset search stage is terminated with the center
having the smallest SAD
in either the first iteration or the second iteration search, the fractional
sample refinement can be
further applied in response.
[0138] As mentioned above, a parametric error surface can be used to derive
the fractional
sample refinement. For example, in a parametric error surface based sub-pixel
offsets estimation,
the center position cost and the costs at four neighboring positions (e.g.,
relative to the center) can
be used to fit a 2D parabolic error surface equation of the following form:
E(x, y) = A (x ¨ xmiii)2 + B (y ¨ y)2 + C
Eq. (2)
[0139] Here, (xmin, ymm) corresponds to the fractional position with the least
cost and C
corresponds to the minimum cost value. By solving Eq. (2) using the cost value
of the five search
points (e.g., the center position and the four neighboring positions), the
(xmin, ymm) can be computed
as:
xmiii = (E(-1,O) ¨ E(1,0))/(2(E(-1,0) + E(1,0) ¨ 2E(0,0)))
Eq. (3)
ymiii = (E(0, ¨1) ¨ E(0,1))/(20(0, ¨1) + E(0,1) ¨ 2E(0,0)))
Eq. (4)
[0140] The values of xmm and ymin can be automatically constrained to be
between -8 and 8 (e.g.,
all cost values are positive and the smallest value is E(0,0)), which can
correspond to a half-pel
offset with 1/16th-pel MV accuracy in VVC. The computed fractional (xmin, ymm)
values can be
added to the integer distance refinement MV (e.g., from the integer sample
offset search described
above) to obtain or determine a sub-pixel accurate refinement delta MV.
[0141] In VVC, motion vector resolution can be 1/16 luma samples. In some
examples, samples
at fractional positions are interpolated using an 8-tap interpolation filter.
In DMVR, the refinement
search points may surround the initial fractional-pel MV with integer sample
offset and a DMVR
search process can include interpolating the samples of the fractional
positions. In some examples,
a bi-linear interpolation filter can be used to generate the fractional
samples for the DMVR search
process. The use of the bi-linear interpolation filter to generate fractional
samples for the DMVR
search process can reduce calculation complexity. In some cases, when a bi-
linear interpolation
41
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
filter is utilized with a 2-sample search range, the DMVR search process may
not need to access
additional reference samples (e.g., relative to existing motion compensation
processes).
[0142] After a refined MV is determined using the DMVR search process, an 8-
tap interpolation
filter can be applied to generate the final prediction. In some examples,
additional reference
samples may not be needed to perform interpolation process based on the
original MV (e.g., as
described above). In some examples, additional samples can be utilized to
perform interpolation
based on the refined MV. Rather than accessing additional reference samples
(e.g., accessing
additional reference samples relative to existing motion compensation
processes), the existing or
available reference samples can be padded to generate additional reference
samples for performing
interpolation based on the refined MV. In some cases, when the width and/or
height of a CU is
larger than 16 luma samples, the DMVR process can further include splitting
the CU can into sub-
blocks each having a width and/or height equal to 16 luma samples, as will be
described in greater
depth below.
[0143] In VVC, DMVR can be applied for CUs that are coded with one or more of
the following
modes and features. In one illustrative example, the modes or features
associated with CUs for
which DMVR can be applied may also be referred to as "DMVR conditions." In
some examples,
the DMVR conditions can include, but are not limited to, CUs that are coded
with one or more of
the following modes and/or features: a CU level merge mode with bi-prediction
MV; one reference
picture in the past (e.g., relative to the current picture) and another
reference picture in the future
(e.g., relative to the current picture); the distances (e.g., POC difference)
from two reference
pictures to the current picture are the same; both reference pictures are
short-term reference
pictures; a CU that includes more than 64 luma samples; both CU height and CU
width are greater
than or equal to 8 luma samples; a BCW weight index indicates equal weight;
weighted prediction
(WP) is not enabled for the current block; combined inter and intra prediction
(CIIP) mode is not
utilized for the current block; etc.
[0144] FIG. 9 illustrates an example extended CU region 900 that can be
utilized to perform bi-
directional optical flow (BDOF) in accordance with some examples of the
disclosure. For example,
BDOF can be used to refine the bi-prediction signal of luma samples in a CU at
a 4x4 sub-block
level (e.g., the 4x4 sub-block 910). In some cases, BDOF can be used for other
sized sub-blocks
42
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
(e.g., 8 x 8 sub-blocks, 4x8 sub-blocks, 8x4 sub-blocks, 16x16 sub-blocks,
and/or other sized sub-
blocks). The BDOF mode can be based on optical flow, which assumes that the
motion of an object
is smooth. As depicted in FIG. 9, the BDOF mode can utilize one extended row
and column around
the boundaries of the extended CU region 900 (e.g., the extended rows 970 and
the extended
columns 980).
[0145] For each 4x4 sub-block (e.g., 4x4 sub-block 910), a motion refinement
(vx, vy) can be
calculated based on minimizing the difference between the LO and Li prediction
samples. The
motion refinement (vx, vy) can then be used to adjust the bi-predicted sample
values in the 4x4 sub-
block (e.g., 4x4 sub-block 910). In one illustrative example, BDOF can be
performed as described
below.
[0146] First, the horizontal and vertical gradients (ai(k)¨ (i,j) and ¨ai(k)
(i,j), respectively, with k
ax ay
= 0,1) of the two prediction signals can be computed by directly calculating
the difference between
two neighboring samples:
¨aloo (i, j) = @M(j + 1,j) >> shift1) ¨ (IM(i ¨ 1,j) >> shift1) Eq. (5)
ax
¨aloo (i,j) = WO(j, j + 1) >> shift1) ¨ (IO(i, j ¨ 1) >> shit t1) Eq. (6)
ay
[0147] Here, /(k)(i,j) represents the sample value at coordinate (i,j) of the
prediction signal in
list k, k = 0,1. shift] can be calculated based on the luma bit depth (e.g.,
bitDepth), as shift] can be
set equal to 6.
[0148] The auto- and cross-correlation of the gradients, Si, S2, S3, S5, and
S6, can subsequently
be calculated as:
= (i,j)En tPx(i,i)
Eq. (7)
Sz = E(i,j)En tPx (0) = sign (iPy(i, l))
Eq. (8)
S3 = E(imEn 61(0) = (¨sign(tpx(i, j)))
Eq. (9)
Ss = E(i,j)En Oy(ii)
Eq. (10)
43
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
S6 = (i men KO) = (¨sign ( ipy(i,j)))
Eq. (11)
[0149] Here,
, aim (i,i) , aim
tPx(i,i) = (- + -(0)) shift3
Eq. (12)
ax a x
H(0) ai
oy(0) = H(0) +m (0)) >> shit t3
Eq. (13)
dY dY
0 (i,j) = (I ( ) (i, j) >> shift2) ¨ (1(1)(0) >> shift2)
Eq. (14)
[0150] Here, Q can be a 6x6 window (or other size window) around the 4x4 sub-
block 910 (or
other sized sub-block). The value of shift2 can be set equal to 4 (or other
suitable value), and the
value of shift3 can be set equal to 1 (or other suitable value).
[0151] The motion refinement (vx, vy) can then be derived using the cross- and
auto-correlation
terms using the following:
võ =S1> 0? clip3 ¨((S3 << 2) >> llog2 J)) : 0
Eq. (15)
vy = s, > 0? c1ip3(¨th'Bio, th'Bro, (((S6 <<2) ¨ ((vi = S2) >> i))>> llog2 Ss
J)): 0 Eq. (16)
[0152] Here, th'13/0=1<<4 and H is the floor function.
[0153] Based on the motion refinement and the gradients, the following
adjustment can be
calculated for each sample in the 4x4 sub-block 910 (or other sized sub-
block):
b(x, y) = v= (a1(1)(x,y) 10)(x,y + v )) (a1(1)(x,y) a I
,y)
)x = Eq. (17)
ax ax ay ay )
[0154] Finally, the BDOF samples of the extended CU region 900 illustrated in
FIG. 9 can be
calculated by adjusting the bi-prediction samples as follows:
Pre d B DO F(x,Y) = (0)) (x, y) + 1(1) (x, y) + b(x, y) + offset) >> shift5
Eq. (18)
44
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0155] Here, shit t5 can be set equal to Max(3, 15 ¨ BitDepth) and the
variable ooffõt can be
set equal to (1 << (shift5 ¨ 1)).
[0156] In some cases, the above values can be selected such that the
multipliers in the BDOF
process (e.g., described above) do not exceed 15-bits, and the maximum bit-
width of the
intermediate parameters in the BDOF process (e.g., described above) is kept
within 32-bits.
[0157] In some examples, the gradient values can be derived based at least in
part on generating
one or more prediction samples i(k)(i,j) in list k where (k= 0,1), wherein the
one or more prediction
samples are outside of the current CU boundaries. As depicted in FIG. 9 and
noted above, BDOF
can utilize one extended row and one extended column around the boundaries of
the extended CU
region 900 (e.g., extended row 970 and extended column 980). In some cases,
the computational
complexity of generating out-of-boundary prediction samples can be controlled
based at least in
part on generating prediction samples in the extended area (e.g., the non-
shaded blocks included
in extended row 970 and extended column 980, along the perimeter of extended
CU region 900)
by taking the reference samples at the nearby integer positions directly
without interpolation. For
example, a floor() operation can be used on the coordinates of the reference
samples at the nearby
reference samples. An 8-tap motion compensation interpolation filter can
subsequently be used to
generate prediction samples within the shaded boxes of extended CU region 900
(e.g., the boxes
internal to the non-shaded perimeter of the extended area prediction samples
of extended row 970
and extended column 980). In some examples, the extended sample values may be
used in gradient
calculation only. Remaining steps of the BDOF process can be performed by
padding (e.g.,
repeating) any sample and gradient values outside of the boundaries of
extended CU region 900
based on their nearest neighbors, as needed.
[0158] In some examples, BDOF can be used to refine the bi-prediction signal
of a CU at the
4x4 sub-block level (or other sub-block levels), as mentioned above. In one
illustrative example,
BDOF can be applied to a CU if the CU satisfies some or all of the following
conditions: the CU
is coded using a "true" bi-prediction mode (e.g., one of the two reference
pictures is located prior
to the current picture in the display order and the other reference picture is
located after the current
picture in the display order); the CU is not coded using affine mode or the
SbTMVP merge mode;
the CU has more than 64 luma samples; both CU height and CU width are greater
than or equal to
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
8 luma samples; BCW weight index indicates equal weight; WP is not enabled for
the current CU;
and/or CIIP mode is not used for the current CU; etc.
[0159] In some aspects, multi-pass decoder side motion refinement can be used.
For example,
in the JVET-V meeting, an enhanced compression model (ECM) was established to
study
compression techniques beyond VVC
(https://vcgit.hhi.fraunhofer.de/ecm/VVCSoftware VTM/-
/tree/ECM). In ECM, a "multi-pass decoder side motion refinement" [JVET U0100]
is adopted to
replace the DMVR in VVC. The multi-pass decoder side motion refinement can
include multiple
passes. In some examples, multi-pass decoder side motion refinement can be
performed by
applying bilateral matching (BM) multiple times. For example, at each pass
(e.g., of the multi-pass
decoder side motion refinement), BM can be applied to a different block size.
[0160] In a first pass, bilateral matching (BM) can be applied to the entire
coding block or coding
unit (e.g., similar to that of DMVR in VVC and/or as described above). For
example, the first pass
BM can be applied to a coding block or CU with a size of 64x64, 64x32, 32x32,
32x16, 16x16,
etc., and combinations thereof.
[0161] In a second pass, BM can again be applied, this time for each 16x16 sub-
block that is
included within (or can be generated from) the overall coding block for which
BM was performed
in the first pass. For example, if the first pass BM was applied to a 64x64
block, the 64x64 block
can be divided into a 4x4 grid of sub-blocks where each sub-block is 16x16. In
the second pass,
BM can be applied to each of the 16 sub-blocks in the example above. In
another example, if the
first pass BM was applied to a 32x32 block, the 32x32 block can be divided
into a 2x2 grid of sub-
blocks that are each 16x16 in size. In this example, the second pass can be
performed by applying
BM to each of the four 16x16 sub-blocks of the 32x32 block. The refined MVs
generated or
obtained from the first pass BM can be utilized as the initial MVs for each
16x16 sub-block for
which the second pass BM is performed.
[0162] In a third pass, a plurality of 8x8 sub-blocks can be obtained or
generated, based on either
the original block (e.g., from the first BM pass) and/or based on the 16x16
sub-blocks (e.g., from
the second BM pass). In some examples, each 16x16 sub-block from the second BM
pass can be
divided into a 2x2 grid of sub-blocks with an 8x8 size. In the third BM pass,
one or more MVs
46
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
associated with each 8x8 sub-block can be further refined by applying bi-
directional optical flow
(BDOF).
[0163] In one illustrative example, multi-pass decoder side motion refinement
can be performed
for a 64x64 block or CU. In a first pass, BM can be applied for the 64x64
block to generate or
obtain a pair of refined MVs. In a second pass, the 64x64 block can be divided
into 16 sub-blocks,
where each sub-block is 16x16 in size. In the second pass, BM can again be
applied, this time to
each of the 16 sub-blocks, and using the refined MVs from the first pass as
the initial MVs. In a
third pass, each 16x16 sub-block can be divided into four sub-blocks with an
8x8 size (e.g., for a
total of 16 * 4 = 64 8x8 sub-blocks for the original 64x64 block). In the
third pass, one or more
MVs associated with each 8x8 sub-block can be refined by applying BDOF, as has
been described
previously.
[0164] Examples of a first pass, second pass, and third pass that can be
included in or utilized to
perform multi-pass decoder side motion refinement are described in turn below.
[0165] In some examples, a first pass can include performing block based
bilateral matching
(BM) motion vector (MV) refinement. For example, in the first pass, a refined
MV is derived by
applying BM to a coding block. Similar to decoder-side motion vector
refinement (DMVR), in the
bi-prediction operation associated with BM a refined MV is searched around two
initial MVs (e.g.,
MVO and MV1) in the reference picture lists LO and Li. The refined MVs (e.g.,
MV0passl and
MV1 pass') are derived around the pair of initial MVs (e.g., MVO and MV1,
respectively) based on
the minimum bilateral matching cost between the two reference blocks in LO and
Ll.
[0166] The first pass BM can include performing a local search to derive
integer sample
precision intDeltallIV. The local search can be performed by applying a 3 x3
square search pattern
(or other search pattern) to loop through the search range [¨sHor, sHor] in a
horizontal direction
and to loop through the search range [¨sVer, sVer] in a vertical direction.
The values of sHor and
sVer can be determined by the block dimension. In some cases, the maximum
value of sHor and/or
sVer can be 8 (or other suitable value).
[0167] The bilateral matching cost can be calculated as:
b//Cost = mvDistanceCost + sadCost
Eq. (19)
47
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0168] When the block size cbW * cbH is greater than 64 (or other block size
threshold), a mean-
removed sum of absolute difference (MRSAD) cost function can be applied to
remove the DC
effect of distortion between reference blocks. When the b//Cost at the center
point of the 3 x3 search
pattern (or other search pattern) has the minimum cost, the intDeltallIV local
search is terminated.
Otherwise, the current minimum cost search point becomes the new center point
of the 3 x3 search
pattern (or other search pattern) and the local search is continued, searching
for the minimum cost
until reaching the end of the search range (e.g., [¨sHor, sHor] in the
horizontal direction and [¨
sVer, sVer] in the vertical direction).
[0169] In some cases, the existing fractional sample refinement can be further
applied to derive
the final deltaMV. The refined MVs after the first pass can then be derived
as:
MV0passl = MVO deltaMV
Eq. (20)
MV1passl = MV1 ¨ deltaMV
Eq. (21)
[0170] In some examples, a second pass can include performing sub-block based
bilateral
matching (BM) motion vector (MV) refinement. For example, in the second pass,
a refined MV
can be derived by applying BM to a 16x16 (or other size) grid sub-block. For
each sub-block, a
refined MV is searched around the two MVs (e.g., MV0passl and MV1passi),
obtained on the first
pass, in the reference picture list LO and Ll.
[0171] Based on the search, two refined MVs, MV0pass2(5bkle2) and
MV1pass2(5b/dx2), can be
derived based on the minimum bilateral matching cost between the two reference
sub-blocks in
LO and Ll. Here, sbidx2 = 0, ..., N-1, is the index for sub-block (e.g.,
because the second pass BM
can be applied to each sub-block generated from the original block used in the
first pass BM). For
example, as described previously, a total of 16 sub-blocks each having
dimension 16x16 can be
generated or obtained for an input block with size 64x64. In this example,
sbidx2 can be an index
to individual ones of the 16 sub-blocks.
[0172] For each sub-block, the second pass BM can include performing a full
search to derive
integer sample precision intDeltallIV. The full search can have a search range
[¨sHor, sHor] in the
horizontal direction and a search range [¨ sVer, sVer] in the vertical
direction. The values of sHor
48
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
and sVer can be determined by the block dimension and the maximum value of
sHor and sVer can
be 8 (or other suitable value).
[0173] FIG. 10 is a diagram illustrating example search area regions in a
coding unit (CU) 1000,
in accordance with some examples of the disclosure. For example, FIG. 10
depicts four different
search regions in the coding unit 1000 (e.g., a first search region 1020, a
second search region
1030, a third search region 1040, a fourth search region 1050, etc.). In some
cases, a bilateral
matching cost can be calculated by applying a cost factor to the SATD cost (or
other cost function)
between two reference sub-blocks, as:
b//Cost = satdCost * costFactor
Eq. (22)
[0174] The search area (2* sHor + 1)* (2* sVer + 1) is divided into five
diamond-shaped search
regions, as depicted in FIG. 10. In other aspects, other search regions can be
used. Each search
region is assigned a costFactor value, which is determined by the distance
(intDeltaiVIV) between
each search point and the starting MV. Each diamond-shaped search region
(e.g., the search
regions 1020, 1030, 1040, and 1050) can be processed in an order starting from
the center of the
search area. Within each search region, search points are processed in the
raster scan order starting
from the top left going to the bottom right corner of the region.
[0175] In some examples, first search region 1020 is searched first, followed
by second search
region 1030, followed by third search region 1040, followed by fourth search
region 1050. When
the minimum b//Cost within the current search region is less than a threshold
equal to sbW * sbH,
the int-pel full search is terminated; otherwise, the int-pel full search
continues to the next search
region until all search points are examined.
[0176] In some cases, the VVC DMVR fractional sample refinement can be further
applied to
derive the final deltaMV(sb/dx2). The refined MVs at the second pass can then
be derived as:
MV0pass2(5b/dx2) = MV0pass 1 deltaMV(sb/dx2)
Eq. (23)
MV1 pass2(sbIdx2) = MV 1 passl ¨ deltaMV(sb/dx2)
Eq. (24)
49
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0177] In some examples, a third pass can include performing sub-block based
bi-directional
optical flow (BDOF) motion vector (MV) refinement. For example, in the third
pass, a refined MV
can be derived by applying BDOF to an 8x8 (or other size) grid sub-block. For
each 8x8 sub-
block, BDOF refinement can be applied to derive a scaled Vx and Vy without
clipping, starting
from the refined MV of the parent sub-block of the second pass. For example,
each parent sub-
block of the second pass may be 16x16 in size (e.g., each parent sub-block of
the second pass can
be associated with four 8x8 sub-blocks used in the third pass). The third pass
BDOF refinement
can be applied or performed based on the refined MVs, MV0pass2(sb/dx2) and
MV1pass2(sb/dx2),
where sbldx2 is an index of one of the parent sub-blocks of the second pass.
[0178] Subsequently, the derived bioMv(Vx, Vy) can be rounded to 1/16 sample
precision and
clipped between -32 and 32 (or other sample precision and/or other clipping
values or ranges).
[0179] The refined MVs (MVOpass3(S X¨
fad 1) and MV1pass3(sbIdx3)) at the third pass can then be
¨
derived as:
MV0pass3(sb/dx3) = MV0pass2(sb/dx2) + bioMv
Eq. (25)
MV I pass3(sbIdx3)= MV0pass2(sb/dx2) ¨ bi oMv
Eq. (26)
[0180] Here, sbldx3 is an index to a particular one of the 8x8 sub-blocks
utilized in the third
pass and sbldx2 is an index to a particular one of the 16x16 sub-blocks
utilized in the second pass.
In some examples, sbldx3 can have a range such that each 8x8 sub-block is
uniquely identifiable
by its sbldx3 index. For example, a 64x64 block can be divided into 64 8x8 sub-
blocks and sbldx3
can contain 64 unique index values for the different 8x8 sub-blocks. In some
examples, sbldx3 can
have a range such that each 8x8 sub-block is uniquely identifiable by its
sbldx3 value in
combination with the sbldx2 value of the corresponding 16x16 parent block. For
example, a 64x64
block can be divided into a total of 16 sub-blocks each having a 16x16 size
and each 16x16 sub-
block can be further divided into four sub-blocks each having an 8x8 size. In
this example, sbldx2
can take one of 16 unique values and sbldx3 can take one of four unique
values, such that each of
the 64 8x8 sub-blocks are identifiable based on the corresponding sbldx2 and
sbldx3 indices.
[0181] Aspects are described herein for improving the search strategies noted
above. Aspects
described herein may be applied to one or more coding (e.g., encoding,
decoding, or combined
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
encoding-decoding) techniques, such as one or more coding techniques in which
a block of a
current picture is predicted from one or more reference pictures (e.g., two
reference blocks from
two respective reference pictures) using motion vectors, where the motion
vectors are refined by
refinement techniques. The improvements may be applied to any suitable video
coding standard
or format (e.g., HEVC, VVC, AV1) as described above, other existing standards
or formats that
apply coding for a block based on reference blocks from two respective
reference pictures and to
any future standards using such techniques. In general, when two reference
blocks from two
reference pictures are used, such techniques are generally referred to as bi-
predicted merge modes
and bilateral matching techniques.
[0182] Rather than following the exact search strategies described above,
various aspects
described herein can allowing different coded blocks to have different search
strategies (or
methods) for bilateral matching. The selected search strategy for a block can
be signaled in one or
more syntax element(s) that is(are) coded in the bitstream. The search
strategy includes the
constraint/relationship between MVDO and MVD1 that is imposed during the
bilateral matching
search process, and may also associate with certain combination of search
pattern, search range or
maximum search rounds, cost criterion etc. In some cases, a constraint can
also be referred to as a
constrain.
[0183] The systems and techniques described herein can be utilized to perform
adaptive bilateral
matching (BM) for decoder side motion vector refinement (DMVR). For example,
the systems and
techniques can perform adaptive bilateral matching for DMVR by applying
different search
strategies and/or search methods for different coded blocks. In some aspects,
a selected search
strategy for a block can be signaled using one or more syntax elements that
are coded in the
bitstream. In some examples, the selected search strategy for a block can be
signaled explicitly,
implicitly, or using a combination thereof As will be described in greater
depth below, a selected
search strategy (e.g., to be used to perform adaptive BM for DMVR for a given
block) can include
a constraint or relationship between MVDO and MVD1, wherein the search
strategy constraint is
utilized or applied during the bilateral matching search process. In some
examples, a search
strategy can additionally, or alternatively, include one or more combinations
of search pattern,
search range, maximum search rounds, cost criterion, etc.
51
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0184] In one illustrative example, the systems and techniques described
herein can perform
adaptive bilateral matching for DMVR using one or more motion vector
difference (MVD)
constraints. As described previously, a motion vector difference can be used
to represent the
difference between an initial motion vector and a refined motion vector (e.g.,
MVDO = MVO' ¨
MVO and MVD1 = MV1' ¨ MV1).
[0185] In some examples, an MVD constraint can be selected (e.g., included in
a selected search
strategy) for a given bilateral matching block. For example, the MVD
constraint can be a mirroring
constraint, in which MVDO and MVD1 have the same magnitude, but opposite sign
(e.g., MVDO
= -MVD1). The MVD mirroring constraint may also be referred to herein as a
"first constraint."
[0186] In another example, the MVD constraint can set MVDO = 0 (e.g., both x
and y
components of MVDO are zero). For example, the MVDO = 0 constraint can be
utilized by holding
MVO as fixed while searching around MV1 to derived refined MV1', and MVO' =
MVO. The
MVDO = 0 constraint may also be referred to herein as a "second constraint."
[0187] In another example, the MVD constraint can set MVD1 = 0 (e.g., both x
and y
components of MVD1 are zero). For example, the mVD1 = 0 constraint can be
utilized by holding
MV1 as fixed while searching around MVO to derive refined MVO', and MV1' =
MV1. The MVD1
= 0 constraint may also be referred to herein as a "third constraint."
[0188] In another example, the MVD constraint can be utilized to search
independently around
MVO to derive MVO' and to search independently around MV1 to derive MV1'. The
MVD
independent search constraint may also be referred to herein as a "fourth
constraint."
[0189] As will be described in greater depth below, in some cases, only the
first and second
constraints may be provided as options (e.g., included in a selected or
signaled search strategy) per
bilateral matching block. In some cases, only the first and third constraints
are provided as options
(e.g., included in a selected or signaled search strategy) per bilateral
matching block. In some
cases, only the first and fourth constraints are provided as options (e.g.,
included in a selected or
signaled search strategy) per bilateral matching block. In some cases, only
the second and third
constraints are provided as options (e.g., included in a selected or signaled
search strategy) per
bilateral matching block. In some cases, the first, second, and third
constraints are provided as
52
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
options (e.g., included in a selected or signaled search strategy) per
bilateral matching block. Any
other combination of the constraints can be provided as options (e.g.,
included in a selected or
signaled search strategy) per bilateral matching block.
[0190] In some aspects, one or more syntax elements can be signaled (e.g., in
a bitstream),
wherein the one or more syntax elements include value(s) to indicate whether
one or more of the
constraints are applied. In some examples, the syntax elements can be used to
indicate or determine
a specific one of the MVD constraints above to apply for a given bilateral
matching block.
[0191] In one illustrative example, a first syntax element can be used to
indicate whether the
first constraint is applied. For example, a first syntax element can be used
to indicate whether the
mirroring constraint (e.g., MVDO = -MVD1) should be applied to a given
bilateral matching block
for which the first syntax element is signaled. In some cases, the first
syntax element can have a
first value when the mirroring constraint is to be applied and a second value
when the mirroring
constraint is not to be applied. In some cases, the presence of the first
syntax element can be used
to infer (e.g., implicitly signal) that the mirroring constraint is to be
applied to the given or current
bilateral matching block, while the absence of the first syntax element can be
used to infer (e.g.,
implicitly signal) that the mirroring constraint should not be applied.
[0192] Continuing in the example above, if the first syntax element indicates
that the first
constraint (e.g., the MVD mirroring constraint, MVDO = -MVD1) is not applied,
a second syntax
element can be used to indicate which of the remaining constraints to apply.
For example, the
second syntax element can be used to indicate whether the second constraint
(e.g., MVDO = 0) or
the third constraint (e.g., MVD1 = 0) should be applied to the given or
current bilateral matching
block for which the syntax elements are signaled. In some examples, the second
syntax element
can have a first value when the second constraint is to be applied and a
second value when the
third constraint is to be applied. In some cases, the presence of the second
syntax element can be
used to infer (e.g., used to implicitly signal) that a pre-determined one of
the second constraint or
the third constraint should be applied, while the absence of the second syntax
element can be used
to infer (e.g., implicitly signal) that the remaining one of the second
constraint or the third
constraint should be applied instead.
53
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0193] In some examples, the one or more syntax elements (e.g., the first
syntax element and/or
the second syntax element) can include mode information, and the selected
constraint from the
aforementioned MVD constraints can be determined based on a mode (e.g., a
merge mode)
indicated by the mode information. For example, the one or more syntax
elements can include
mode information that indicates different merge modes for the current
bilateral matching block. In
one illustrative example, the first constraint (e.g., the mirroring
constraint, MVDO = -MVD1) can
be applied to a regular (e.g., standard or default, such as extended merge
prediction in VVC merge
mode) merge mode coded block when a regular merge candidate satisfies the DMVR
conditions
noted above.
[0194] As described previously, the DMVR conditions can indicate CUs that are
coded with one
or more of the following modes and features. In some examples, the DMVR
conditions can
include, but are not limited to, CUs that are coded with one or more of the
following modes and/or
features: a CU level merge mode with bi-prediction MV; one reference picture
in the past (e.g.,
relative to the current picture) and another reference picture in the future
(e.g., relative to the
current picture); the distances (e.g., POC difference) from two reference
pictures to the current
picture are the same; both reference pictures are short-term reference
pictures; a CU that includes
more than 64 luma samples; both CU height and CU width are greater than or
equal to 8 luma
samples; a BCW weight index indicates equal weight; weighted prediction (WP)
is not enabled for
the current block; combined inter and intra prediction (CIIP) mode is not
utilized for the current
block; etc.
[0195] In another illustrative example, one of the second constraint (e.g.,
MVDO = 0) or the third
constraint (e.g., MVD1 = 0) can be applied when the coded block uses a
designated new merge
mode (e.g., an adaptive bilateral matching mode, as described herein) and
wherein all merge
candidates meet the DMVR conditions. In some cases, the second constraint
and/or the third
constraint additionally or alternatively be indicated by a mode flag or merge
index. For example,
an indication of or selection between the second constraint and the third
constraint may be
determined based on a mode flag or merge index.
[0196] In some examples, the one or more syntax elements described herein can
include an index
of a merge candidate list. The selected constraint can then be determined by
the index that indicates
54
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
the selected merge candidate from a merge candidate list (e.g., the constraint
depends on the
selected merge candidate). In still another example, the syntax elements can
include a combination
of mode flags and indices.
[0197] In some aspects, the systems and techniques described herein can signal
(e.g., explicitly
and/or implicitly) a selected search strategy that can be utilized to perform
the multi-level (e.g.,
multi-pass) DMVR described above. In some examples, a selected search strategy
can be applied
in one pass of the multi-pass DMVR. In another example, a selected search
strategy can be applied
in multiple levels or passes of the multi-pass DMVR (e.g., but not all levels
or passes of the process
in some cases).
[0198] For the example of the three-pass DMVR that is described above, in one
illustrative
example a selected strategy may only be applied in the first pass (e.g., the
PU-level bilateral
matching). The second and third passes (e.g., which perform bilateral matching
for a first sub-
block size and BDOF for a second sub-block size smaller than the first sub-
block size, respectively)
can be performed utilizing a default strategy, for example, those described
above with respect to
FIG. 10 and a standardized three pass structure. In one illustrative example,
multi-pass DMVR
(e.g., the three-pass DMVR described above) can utilize the second search
strategy (e.g., the
second constraint MVDO = 0) and/or the third search strategy (e.g., the third
constraint MVD1 =
0) in the first pass to perform PU-level bilateral matching. Subsequent passes
(e.g., the second and
third passes) can use a default search strategy that includes the first
constraint (e.g., the mirroring
constraint MVDO = -MVD1).
[0199] In some aspects, the search strategies can be grouped into multiple
subsets. In some
examples, one or more syntax elements can be used to determine the selected
subsets. In some
cases, a selected strategy within a given subset may be determined implicitly.
In one illustrative
example, the first constraint (e.g., the mirroring constraint MVDO = - MVD1)
can be included in
a first subset and the second constraint (e.g., MVDO = 0) and the third
constraint (e.g., MVD1 =
0) can both be included in a second subset. A syntax element can be used to
indicate whether the
second subset is used. If the second subset is used (e.g., based on the
corresponding syntax
element), then a selection or determination between applying the second
constraint or the third
constraint can be determined implicitly. For example, an implicit
determination can be made based
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
on the minimum matching cost between using the second constraint for bilateral
matching and
using the third constraint for bilateral matching. If bilateral matching using
the second constraint
is determined to provide a smaller matching cost than bilateral matching using
the third constraint,
then the second constraint can be selected, otherwise the third constraint is
selected.
[0200] In some examples, one or more aspects of the systems and techniques
described herein
can be utilized with or applied based on the enhanced compression model (ECM).
For instance, in
ECM, multi-pass DMVR can be applied to a regular (e.g., a standard or default,
such as extended
merge prediction in VVC merge mode) merge mode candidate with one or more
features that are
the same as or similar to those described above. For example, the one or more
features can be the
same as or similar to some (or all) of the DMVR conditions described above. As
noted above, a
merge candidate refers to a candidate block from which information (e.g., one
or more motion
vectors, prediction mode, etc.) is inherited for use in coding (e.g., encoding
and/or decoding) a
current block, where the candidate block may be a block neighboring the
current block. For
instance, a merge candidate may be an inter-predicted PU that includes a
motion data position
selected from a group of spatially neighboring motion data positions and one
of two temporally
co-located motion data positions.
[0201] Multi-pass DMVR in ECM can be performed based on applying the first
constraint (e.g.,
the mirroring constraint MVDO = -MVD1) by default. In one illustrative
example, the systems and
techniques described herein can utilize an adaptive bilateral matching mode as
a new mode for the
multi-pass DMVR. In some examples, the adaptive bilateral matching mode can
also be referred
to as "adaptive bm mode." In the adaptive bm mode, a merge index can be
signaled to indicate
the selected motion information candidate. However, all the candidates in the
candidate list can
satisfy the DMVR conditions. In some cases, a flag (e.g., a bm merge flag) can
be used to signal
or indicate the use of the adaptive bilateral matching mode. For example, if
the flag is true (e.g.,
bm merge flag is equal to 1), adaptive bm mode can be used or applied (e.g.,
as will be described
below). In some aspects, when the flag is true (e.g., bm merge flag is equal
to 1), an additional
flag (e.g., bm dir flag) can be used to signal or indicate a bm dir value to
be used in the
adaptive bm mode.
56
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0202] In some aspects, if adaptive bm mode = 1, adaptative bilateral matching
can be
performed, and if adaptive bm mode = 0, then adaptive bilateral matching is
not performed. In
one illustrative example, an adaptive bilateral matching process (e.g.,
associated with
adaptive bm mode = 1) can be performed based at least in part on applying
either the second
constraint or the third constraint to the selected candidate (e.g., fixing
either MVDO or MVD1 as
equal to 0, respectively). In some cases, a variable bm dir may be used to
indicate which constraint
is applied and/or to indicate a selected constraint. For example, when
adaptive bilateral matching
is performed (e.g., signaled or determined based on adaptive bm mode), bm dir
= 1 can be used
to indicate or signal that adaptive bilateral matching should be performed by
fixing MVD1 to 0
(e.g., the third constraint should be applied). In some examples, bm dir = 2
can be used to indicate
or signal that adaptive bilateral matching should be performed by fixing MVDO
to 0 (e.g., the
second constraint should be applied).
[0203] In some cases, if adaptive bilateral matching is not performed, a
regular merge mode can
be utilized. As mentioned previously, regular merge mode can be determined or
signaled based on
adaptive bm mode = 0, which indicates that adaptive bilateral matching is not
performed. In some
examples, when regular merge mode is utilized (e.g., because adaptive
bilateral matching is not
performed), the systems and techniques can utilize an inferred value of bm dir
= 3, which indicates
or signals that MVDO and MVD1 are not fixed and that MVDO = -MVD1 (e.g., the
first, mirroring
constraint should be applied). In some examples, bm dir = 3 can be explicitly
signaled or used to
indicate regular merge mode.
[0204] In some examples, the systems and techniques can perform bilateral
matching with one
or more modified bilateral matching operations. In some aspects, when bm dir =
3, the bilateral
matching process can be the same as the three-pass bilateral matching process
previously described
above. For example, given an initial pair of MVs, a first predictor is
generated by the first MV
referencing to the first reference picture and a second predictor is generated
by the second MV
referencing to the second reference picture. A refined pair of MVs are
subsequently derived by
minimizing the BM cost between a refined first predictor (e.g., generated
using the first refined
MV) and a refined second predictor (e.g., generated using the second refined
MV), wherein the
motion vector differences between the refined MVs and initial MVs are MVDO and
MVD1, and
MVDO = - MVD 1 .
57
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0205] In some aspects, when bm dir = 1, only the first MV is refined, while
the second MV is
held fixed. For example, the refined first MV can be derived by minimizing the
BM cost between
a second predictor that is generated by the second MV and a refined first
predictor that is generated
by the refined first MV. For example, when bm dir = 1, MVO can be refined
while MV1 is held
fixed. A refined motion vector, MVO', can be derived by minimizing the BM cost
between a
predictor generated based on MV1 and a refined predictor that is generated
based on MVO'.
[0206] In some examples, when bm dir = 2, only the second MV is refined, while
the first MV
is held fixed. For example, the refined second MV can be derived by minimizing
the BM cost
between a first predictor that is generated by the first MV and a refined
second predictor that is
generated by the refined second MV. For instance, when bm dir = 2, MV1 can be
refined while
MVO is held fixed. A refined motion vector, MV1', can be derived by minimizing
the BM cost
between a predictor generated based on MV1 and a refined predictor that is
generated based on
MV1'.
[0207] In some aspects, a BM cost (e.g., as described above) can additionally,
or alternatively,
include a regularization term based on or derived from a motion vector
difference (MVD). In one
illustrative example, a multi-pass DMVR search process can include a
regularization term
determined based on an MV cost that depends on the refined MV position.
[0208] In some aspects, one or more of the bilateral matching modifications
described above
may be applied only in the first pass of a multi-pass bilateral matching
process (e.g., the PU-level
DMVR). In other aspects, one or more of the bilateral matching modifications
described above
can be applied in both a first pass and a second pass (e.g., applied in the PU-
level and the sub-PU-
level DMVR passes).
[0209] In some examples, the systems and techniques described herein can
perform multi-pass
bilateral matching DMVR using a greater or lesser quantity of passes than
described in the
examples above. For instance, fewer than three passes could be utilized and/or
greater than three
pass could be utilized. In some examples, any number of passes may be used,
and the passes may
be structured in any fashion. In some aspects, the multi-pass design can be
similarly applied in
adaptive bm mode. In some aspects, the second pass may be skipped in adaptive
bm mode. In
some aspects, the second pass and the third pass may both be skipped in
adaptive bm mode. In
58
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
other aspects, any combination of passes (e.g., pass operations from the three-
pass system
described above) can be combined with repeated passes or other pass types
based on particular
adaptive bilateral matching criteria.
[0210] In some aspects of multi-pass DMVR, different search patterns can be
used for different
search levels and/or search precision(s). For example, a square search can
used for both the integer
search and the half-pel search that may be applied to perform PU-level DMVR
(e.g., the first pass).
In some examples, for sub-PU-level DMVR (e.g., the second pass and/or third
pass), full search
can be used for the integer search and square search can be used for the half-
pel search. In some
examples, one or more (or all) of the search patterns described above can be
utilized based on a
determination that bm dir is equal to a first (a value of 1), second (a value
of 2), or third value (a
value of 3). In one example, one or more (or all) of the search patterns
described above can be
utilized based on a determination that bm dir = 3.
[0211] The following aspects describe example search patterns and/or search
process when
bm dir equals 1 or 2. In one aspect, the same search patterns as when bm dir =
3 can be used when
bm dir = 1 and/or when bm dir =2. In another aspect, bm dir = 1 and bm dir = 2
can use different
search patterns than when bm dir = 3. For instance, in one example, full
search can be used for
the integer search in PU-level DMVR and square search can be used for the half-
pel search in PU-
level DMVR.
[0212] In some aspects, the same search range and/or maximum quantity of
search rounds can
be used for the different values of bm dir. In other aspects, when bm dir = 1
or 2, a different
search range and/or a different maximum quantity of search rounds can be used.
For example, in
the case of full search with different cost factors assigned to different
MVDs, one or more MVD
regions may be skipped. For example, as illustrated in FIG. 10, the search
area for CU 1000 is
divided into multiple search regions (e.g., first search region 1020, second
search region 1030,
third search region 1040, fourth search region 1050). In some cases, regions
that are farther from
the center of the search area (e.g., farther from first search region 1020)
may be skipped.
[0213] In some examples, in regular merge mode all search regions may be
searched. For
example, with respect to FIG. 10, in regular merge mode the four search
regions 1020, 1030, 1040,
and 1050 can be searched, along with a fifth search region that includes the
remaining blocks of
59
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
CU 1000 that are not already included in one of the four search regions 1020-
1050. As mentioned
above, in some cases search regions that are farther from the center of the
search area can be
skipped. For example, in adaptive bilateral matching mode (e.g., when bm dir =
1 or 2), only the
first three search regions associated with the CU 1000 of FIG. 10 may be
searched (e.g., in adaptive
bilateral matching mode, only first search region 1020, second search region
1030, and third search
region 1040 may be searched).
[0214] In some aspects of multi-pass DMVR, the SAD or mean removal SAD (e.g.,
depending
on PU size) can be used for the integer search and the half-pel search
associated with a PU-level
DMVR pass. In some cases, SATD can be used for the sub-PU-level DMVR. In some
aspects, the
same cost criterions can be used for different values of bm dir. For example,
the current cost
criterions selection in ECM can be used for all values of bm dir. In other
aspects, cost criterions
selection may be different for different values of bm dir. For example, when
bm dir = 3, the cost
criterions selection in current ECM can be applied. The SAD or mean removal
SAD, depending
on the PU size, can be used for integer and half-pel search in PU-level DMVR.
When bm dir = 1
or 2, the PU-level DMVR process can use SATD for the integer search and can
use SAD for the
half-pel search.
[0215] As mentioned previously, candidates in the candidate list for some
aspects satisfy the
DMVR conditions. In one additional aspect, bm dir can be set equal to 1 or 2
as indicated by an
additional bm dir flag included in the adaptive bm mode mode. In some cases,
the candidate list
for adaptive bm mode is generated on top of the regular merge candidate list.
For example, one
or more candidates that are determined to satisfy the DMVR conditions in the
regular merge
candidate list can be inserted into the candidate list for adaptive bm mode.
[0216] In another additional aspect, whether to set bm dir equal to 1 or 2 can
be indicated by
the merge index within the adaptive bm mode mode. The candidate list for
adaptive bm mode
can be generated on top of regular merge candidate list. For each candidate in
the regular merge
candidate list that satisfies the DMVR conditions, a pair of two candidates
can be inserted into the
adaptive bm mode candidate list, one candidate with bm dir = 1 and the other
candidate with
bm dir = 2, wherein the two candidates in the pair have identical motion
information. In some
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
examples, the bm dir can be determined by determining whether the merge index
is an even or
odd number.
[0217] In one illustrative example, the candidate list for adaptive bm mode
can be generated
independently from the regular merge candidate list. In some cases, generation
of the candidate
list for adaptive bm mode can follow the same or similar process as generating
the candidate list
for the regular merge mode (e.g., checking the same spatial, temporal
neighboring positions,
history-based candidates, pair-wise candidates, etc.). In some cases, pruning
may be applied during
the list construction process.
[0218] In still another aspect, one or more candidates associated with a bi-
prediction with CU-
level weight (BCW) weight index that indicates non-equal weight may also be
added into the
candidate list (e.g., compared to some systems with DMVR conditions, where BCW
weight index
can indicate equal weight).
[0219] In some examples, a padding processing may be applied if the number of
candidates in
the candidate list for adaptive bm mode and/or in the candidate list for
regular merge mode is less
than a pre-defined maximum. For instance, once the candidate list for adaptive
bm mode is
generated, there may be fewer candidates in the list than a pre-defined
maximum number of
candidates. In such an example, the padding process can be applied to generate
a number of padded
candidates for the candidate list so that the padded candidate list includes
the pre-defined number
of candidates. In one illustrative example, in the case that adaptive
bilateral matching is enabled
(e.g., adaptive bm mode = 1), one or more default candidates can be used for
padding in the merge
list construction. A default candidate may be derived such that it satisfies
the DMVR conditions.
[0220] In some examples, the MVs may be set to zero for the default
candidates. For instance,
zero MV candidates can be added during the padding process. In some examples,
the reference
pictures can be selected according to the DMVR conditions. In some cases, the
reference index
can loop over all possible values until the number of candidates in the
candidate list reaches the
maximum number of candidates (e.g., the pre-defined maximum). In another
aspect, BCW weight
index can be used to indicate equal weight for the regular candidates, and one
or more non-equal
weight BCW candidates can be added thereafter and before adding zero
candidates.
61
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0221] In some aspects, the reference pictures assigned to default candidates
(e.g., the reference
pictures assigned to the padded zero MV candidates described above) can be
selected to satisfy
one or more conditions associated with the adaptive bm mode. An illustrative
example of such
conditions may include one or multiple of the following conditions and/or may
include other
conditions not listed here: at least one pair of reference pictures is
selected that includes one
reference picture in the past and one reference picture in the future,
relative to the current picture;
the respective distances from both reference pictures to the current picture
are equal; both of the
reference pictures are not long-term reference pictures; both of the reference
pictures have the
same resolution as the current picture; weighted prediction (WP) is not
applied to any of the
reference pictures; any combination thereof; and/or other conditions.
[0222] In some aspects, one or more reference pictures can be assigned to the
default candidates
(e.g., to the padded zero MV candidates described above) based on selecting
reference pictures
that satisfy one or more conditions associated with or based on a specified or
selected constraint
(e.g., the second constraint, MVDO = 0, or the third constraint, MVD1 = 0). In
some examples,
reference pictures can be selected to satisfy one or more conditions
associated with a given
constraint, wherein the given constraint is determined based on bm dir (e.g.,
as described
previously). For example, the one or more conditions based on bm dir may be
applied only to the
MV for which refinement is performed (e.g., if bm dir = 1). An illustrative
example of such
conditions may include one or multiple of the following and/or may include
other conditions not
listed here: the reference picture in a given List X is not a long-term
reference picture; the reference
picture in List X has the same resolution as the current picture; weighted
prediction (WP) is not
applied to the reference picture in List X; the respective distance from a
first reference picture in
List X to the current picture is not smaller than the respective distance from
the other reference
picture (e.g., a second reference picture in list X) to the current picture;
any combination thereof;
and/or other conditions.
[0223] In some cases, list X can be the same as List 0 (e.g., list LO) if bm
dir indicates an MV
in List 0 is refined. In some cases, list X can be the same as List 1 (e.g.,
list L1) if bm dir indicates
an MV in List 1 is refined. In some aspects, all possible zero MV candidates
can be found by
looping over all possible combinations of reference pictures and identifying
the reference pictures
that satisfy the pre-defined conditions in a certain order. In one
illustrative example, a first loop
62
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
can be performed for List 0 and a second loop can be performed for List 1. In
another example,
the first loop can be performed for List 1 and the second loop can be
performed for List 0. Other
ordering is also possible and shall be considered within the scope of the
present disclosure. The
process of determining the possible zero MV candidates (e.g., by looping over
the combinations
of reference pictures to identify reference pictures that satisfy the pre-
defined conditions in a
certain order) may be performed at the slice level, picture level, or other
level. The list of the
identified default MV candidates (e.g., the zero MV candidates) can be stored
as default
candidates. In some cases, when determining the possible zero MV candidates at
the block level,
when the number of candidates is less than the pre-defined maximum number of
candidates, the
systems and techniques described herein may loop over the default candidates
and add one or more
of the default candidates into the candidate list until the number of
candidates reaches the pre-
defined maximum.
[0224] In some aspects, one or more size constraints can be included in and/or
utilized by the
adaptive bm mode described herein. In one aspect, the same size constraint as
in regular DMVR
can be applied in adaptive bm mode. In another aspect, the adaptive bm mode is
not applied if
neither the width nor the height of the current block is larger than the
minimum block size for
DMVR.
[0225] In some aspects, the adaptive bm mode can be signaled as an additional
merge mode to
the regular merge mode. In some examples, various signaling methods may be
applied or utilized
to signal adaptive bm mode as an additional merge mode. For example, the
adaptive bm mode
can be considered or signaled as a variant of regular merge mode. In one
illustrative examples, one
or more syntax elements can first be signaled to indicate the regular merge
mode, and one or more
additional flags and/or syntax elements can be signaled to indicate the
adaptive bm mode and/or
to indicate a specific one of the constraints (e.g., the second constraint
MDVO = 0 or the third
constraint MDV1 = 0) that may be applied in associated with use of the
adaptive bm mode.
[0226] In another aspect, adaptive bm mode can be indicated by one or more
flags prior to the
indication of regular merge mode. For example, if the syntax (e.g., one or
more first syntax
elements) indicates the current block is not using adaptive bm mode, then one
or more additional
syntax elements may be signaled to indicate whether the current block uses
regular merge mode
63
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
or other merge modes. For example, if the current block does not use adaptive
bm mode or regular
merge mode, the one or more additional syntax elements may be signaled
indicate that the current
block uses other merge modes such as combined inter and intra prediction
(CIIP), geometric
partition mode (GPM), etc.
[0227] In still another aspect, adaptive bm mode may be signaled in other
merge mode
branches. For example, adaptive bm mode can be signaled in a template matching
merge mode
branch. In some cases, one or more syntax elements may first be signaled to
indicate if one of the
adaptive bm mode and the template matching merge mode is used. If the one or
more syntax
elements indicate that adaptive bm mode or template matching merge mode is
used, then one or
more additional flags or syntax elements can be signaled to indicate which one
of template
matching merge mode and adaptive bm mode is used.
[0228] In some aspects, the merge index in adaptive bm mode can use the same
signaling
method as in regular merge mode. In one aspect, adaptive bm mode can use the
same (or similar)
context models as in regular merge mode. In another aspect, separate context
models may be used
for adaptive bm mode. In some examples, the maximum number of merge candidates
may be
different for adaptive bm mode than the maximum number of merge candidates for
regular merge
mode.
[0229] In one illustrative example, one or more high-level syntax elements may
be used to
indicate whether adaptive bm mode can be applied or is to be applied. In one
aspect, the same
high-level syntax that is used to indicate whether regular DMVR is to be
applied may also be used
to indicate whether adaptive bm mode is to be applied. In another aspect, one
or more separate
(e.g., additional) high-level syntax elements can be used to indicate whether
adaptive bm mode
should be applied. In still another aspect, separate high-level syntax
elements may be used to
indicate whether adaptive bm mode is utilized, wherein the separate high-level
syntax elements
for adaptive bm mode are present only if regular DMVR is enabled. For example,
if the separate
high-level syntax elements associated with regular DMVR are determined to
indicate that the
regular DMVR is not enabled or utilized, then the separate or additional high-
level syntax elements
associated with adaptive bm mode are not signaled and adaptive bm mode is
inferred to be off
(e.g., not enabled or utilized).
64
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0230] In some aspects, in addition to the one or more high-level syntax
elements described
above, adaptive bm mode can be disabled for a coded picture or slice according
to available
reference pictures. In some cases, if it is determined that no combination of
reference pictures
satisfies or can satisfy the reference picture conditions, then the adaptive
bm mode can be
disabled and the corresponding syntax elements (e.g., at the block level) are
not signaled. In some
cases, there must be at least one pair of reference pictures that satisfy the
reference picture
conditions in order to utilize adaptive bm mode. An illustrative example of
such conditions may
include one or multiple of the following and/or may include other conditions
not listed here: one
reference picture in the past and one reference picture in the future
,relative to the current picture;
the respective distances from both reference pictures to the current picture
are equal; both reference
pictures are not long-term reference pictures; both reference pictures have
the same resolution as
the current picture; weighted prediction (WP) is not applied to either
reference pictures; any
combination thereof; and/or other conditions.
[0231] The conditions listed above can be used separately or in a combination.
[0232] In some aspects, only a subset of the adaptive bilateral matching modes
described herein
(e.g., a first adaptive bilateral matching mode associated with the second
condition MDVO = 0,
and a second adaptive bilateral matching mode associated with the third
condition MDV1 = 0)
may be enabled depending on the reference pictures. An illustrative example of
such conditions to
only allow bm dir = 1 (e.g., associated with the second condition MDVO = 0) or
bm dir = 2 (e.g.,
associated with the third condition MDV1 = 0) may include one or multiple of
the following and/or
may include other conditions not listed here: one of the reference pictures is
long-term reference
picture and the other reference picture is not long-term reference picture;
one of the reference
pictures has the same resolution as the current picture but the other
reference picture has a different
resolution than the current picture; weighted prediction (WP) is applied to
one of the reference
pictures; the respective distance from one of the reference pictures to the
current picture is not
shorter than the respective distance from the other reference picture to the
current picture; any
combination thereof; and/or other conditions.
[0233] In some cases, a syntax element (e.g., at the block level) that
identifies the bilateral
matching mode may not be signaled and can inferred accordingly. In some
examples, the syntax
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
element can be implicitly signaled (e.g., inferred) rather than being
explicitly signaled (e.g., in a
bitstream as part of a particular syntax table). In some cases, the syntax
element may be neither
explicitly signaled nor implicitly signaled, and can be inferred. For example,
if bm dir of a first
value is enabled (e.g., bm dir = 1 is enabled) but the bm dir of a second
value is disabled (e.g.,
bm dir = 2 is disabled), then a syntax element used to indicate bm dir at the
block level may not
signaled. In the absence of this syntax element, bm dir can be inferred to be
a first value (e.g.,
bm dir is inferred to be 1). In another example, if bm dir of a second value
is enabled (e.g., bm dir
= 2) is enabled but bm dir of a first value is disabled (e.g., bm dir = 1 is
disabled), then the syntax
element used to indicate bm dir at the block level may not be signaled. In the
absence of this
syntax element, bm dir can be inferred to be a second value (e.g., bm dir is
inferred to be 2).
[0234] In another example, a slice level flag and/or a picture level flag can
be utilized for
adaptive bm mode. For example, a slice level flag and/or a picture level flag
can be utilized as a
bitstream conformance constraint that the flag is set to 0 (e.g., DMVR modes
are disabled) if at
least one of the above conditions is not satisfied.
[0235] In yet another example, a bitstream conformance constraint can be
introduced to the
existing signaling, wherein the bitstream conformance constraint indicates
that the
adaptive bm mode should not be applied and the corresponding overhead set to 0
(e.g., indicating
that the adaptive bm mode is not used), if at least one of the above
conditions is not satisfied.
[0236] In some aspects of ECM, multiple hypothesis prediction (MHP) can be
utilized. In MHP,
an inter prediction technique can be used to obtain or determine a weighted
superposition of more
than two motion-compensated prediction signals. The resulting overall
prediction signal can be
obtained based on performing sample-wise weighted superposition. For example,
based on a uni/bi
prediction signal puma,' and a first additional inter prediction
signal/hypothesis h3, the resulting
prediction signal p3 can be obtained as follows:
P3 = (1 - a)Punimi ah3
Eq. (27)
[0237] Here, the weighting factor a can be specified by a syntax element add
hyp weight idx,
according to the following mapping:
66
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
add_hyp_weight_idx a
0 1/4
1 -1/8
Table 2
[0238] In some examples, more than one additional prediction signals can be
used. In some
cases, the more than one additional prediction signals can be utilized in a
manner that is the same
as or similar to the above. For example, when utilizing multiple additional
prediction signals, a
resulting overall prediction signal can be accumulated iteratively with each
additional prediction
signal as follows:
Pn+1 = (1 ¨ an+t)Pn an-E1hn+1
Eq. (28)
[0239] Here, the resulting overall prediction signal can be obtained as the
last pn (e.g., the p
having the largest index n) .
[0240] In some aspects, MEW may not be applied (e.g., is disabled) for any
adaptive bm mode.
In some aspects, MEW may be applied on top of adaptive bm mode in the same or
similar manner
as MEW is applied for a standardized (e.g., regular) merge mode.
[0241] FIG. 11 is a flowchart illustrating an example of a process 1100 for
processing video
data. In some examples, process 100 can be used to perform decoder side motion
vector refinement
(DMVR) with adaptive bilateral matching, in accordance with some examples of
the disclosure.
In some aspects, process 1100 can be implemented in an apparatus for
processing video data
comprising a memory and one or more processors coupled to the memory
configured to perform
the operations of process 1100. In other aspects, process 1100 can be
implemented in non-
transitory computer readable medium comprising instructions that, when
executed by one or more
processors of a device, cause the device to perform the operations of process
1100.
[0242] At block 1102, the process 1100 includes obtaining one or more
reference pictures for a
current picture (e.g., a current block of the current picture). For example,
the one or more reference
pictures can be obtained based on one or more of the inputs 114 to decoding
device 112 illustrated
in FIG. 1. In some examples, the one or more reference pictures and the
current pictures can be
obtained from a video data obtained by or provided to decoding device 112
illustrated in FIG. 1.
67
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0243] At block 1104, the process 1100 includes identifying a first motion
vector and a second
motion vector for a merge mode candidate. For example, the first motion vector
and/or the second
motion vector can be identified by decoding device 112 illustrated in FIG. 1.
In some examples,
the first motion vector and/or the second motion vector can be identified
using the decoder engine
116 of decoding device 112 illustrated in FIG. 1. In some cases, one or more
(or both) of the first
motion vector and the second motion vector can be identified using signaled
information. For
example, encoding device 104 illustrated in FIG. 1 can include signaling
information that can be
used by decoding device 112 and/or decoding engine 116 to identify one or more
(or both) of the
first motion vector and the second motion vector. In some cases, the process
1100 may include
determining a merge mode candidate for the current picture. As noted herein,
the merge mode
candidate may include a neighboring block of the block from which prediction
data can be
inherited for a block of the current picture. For example, the merge mode
candidate can be
determined by decoding device 112 illustrated in FIG. 1. In some examples, the
merge mode
candidate can be determined using the decoder engine 116 of decoding device
112 illustrated in
FIG. 1. In some examples, signaling information can be used by decoding device
112 and/or
decoding engine 116 to determine the merge mode candidate for the current
picture. In some cases,
the merge mode candidate can be determined used the same signaling information
used to identify
one or more (or both) of the first motion vector and the second motion vector
associated with the
merge mode candidate. In some cases, the merge mode candidate and the first
and second motion
vectors can be determined using separate signaling information.
[0244] At block 1106, the process 1100 includes determining a selected motion
vector search
strategy for the merge mode candidate from a plurality of motion vector search
strategies. In some
aspects, the selected motion vector search strategy is associated with one or
more constraints based
on or corresponding to the first motion vector and/or the second motion
vector. In one illustrative
example, the selected motion vector search strategy can be a bilateral
matching (BM) motion
vector search strategy. In some cases, the motion vector search strategy for
the merge mode
candidate can be selected from a plurality of motion vector search strategies
that include at least
two of a multi-pass decoder side motion vector refinement strategy, a
fractional sample refinement
strategy, a bi-directional optical flow strategy, or a sub-block based
bilateral matching motion
vector refinement strategy. In some examples, the selected motion vector
search strategy can be
determined before the merge mode candidate is determined. For example, the
selected motion
68
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
vector search strategy may be determined (e.g., as described above) and used
to generate a merge
candidate list. The merge mode candidate can be determined based on a
selection from the
generated merge candidate list. In some examples, the selected merge mode
candidate may be
determined before the selected motion vector search strategy is determined.
For example, in some
cases the merge candidate list may be generated without using the selected
search strategy (e.g.,
the generated merge candidate list may be the same for each respective search
strategy of the
plurality of search strategies), and the selected merge candidate can be
determined before the
selected search strategy.
[0245] In some examples, the selected motion vector search strategy can be a
multi-pass decoder
side motion vector refinement (DMVR) search strategy. For example, a multi-
pass DMVR search
strategy can include one or more block based bilateral matching motion vector
refinement passes
and can also include one or more sub-block based motion vector refinement
passes. In some
examples, the one or more block based bilateral matching motion vector
refinement passes can be
performed using a first constraint that is associated with a first motion
vector difference and/or a
second motion vector difference. The first motion vector difference can be a
difference determined
between the first motion vector and a refined first motion vector. The second
motion vector
difference can be a difference determined between the second motion vector and
a refined second
motion vector. In some examples, the one or more sub-block based motion vector
refinement
passes can be performed using a second constraint that is different than the
first constraint. The
second constraint can be associated with at least one of the first motion
vector difference and/or
the second motion vector difference, as described above. In some cases, the
one or more sub-block
based motion vector refinement passes can include at least one of a sub-block
based bilateral
matching (BM) motion vector refinement pass and/or a sub-block based bi-
directional optical flow
(BDOF) motion vector refinement pass.
[0246] In some examples, the selected motion vector search strategy is
associated with one or
more constraints corresponding to at least one of the first motion vector or
the second motion
vector (e.g., as described above). The one or more constraints can be
determined based on one or
more signaled syntax elements. For instance, the one or more constraints can
be determined for
the block of the current picture based on a syntax element signaled for the
block. In some aspects,
the one or more constraints are associated with at least one of a first motion
vector difference
69
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
associated with the first motion vector (e.g., the difference between the
first motion vector and the
refined first motion vector) and a second motion vector difference associated
with the second
motion vector (e.g., the difference between the second motion vector and the
refined second
motion vector). In some examples, the one or more constraints can include a
mirroring constraint
for the first motion vector difference and the second motion vector
difference. The mirroring
constraint can set the first motion vector difference and the second motion
vector difference to
have an equal magnitude (e.g., absolute value) but an opposite sign. In some
cases, the one or more
constraints can include a zero value constraint for the first motion vector
difference (e.g., setting
the first motion vector difference equal to zero). In some examples, the one
or more constraints
can include a zero value constraint for the second motion vector difference
(e.g., setting the second
motion vector difference equal to zero). In some aspects, the zero value
constraint can be indicative
of maintaining a motion vector difference as fixed. For instance, based on the
zero value constraint,
the process 1100 may include determining the one or more refined motion
vectors using the
selected motion vector search strategy by maintaining a first one of the first
motion vector
difference or the second motion vector difference as a fixed value and
searching relative to a
second one of the first motion vector difference or the second motion vector
difference. For
instance, the first motion vector difference can be fixed and a search can be
performed around the
second motion vector difference to derive a refined motion vector.
[0247] At block 1108, the process 1100 includes determining, using the
selected motion vector
search strategy, one or more refined motion vectors based on the first motion
vector, the second
motion vector, and/or the one or more reference pictures (e.g., based the
first motion vector and
the one or more reference pictures, based the second motion vector and the one
or more reference
pictures, or based the first motion vector, the second motion vector, and the
one or more reference
pictures). In some cases, determining the one or more refined motion vectors
may include
determining the one or more refined motion vectors for a block of the video
data. In some
examples, the one or more refined motion vectors can include a first refined
motion vector and a
second refined motion vector, determined for the first motion vector and the
second motion vector,
respectively. In some examples, the first motion vector difference is
determined as a difference
between the first refined motion vector and the first motion vector, and the
second motion vector
difference is determined as a difference between the second refined motion
vector and the second
motion vector.
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0248] In some examples, the selected motion vector search strategy is a
bilateral matching (BM)
motion vector search strategy, as mentioned previously. When the selected
motion vector search
strategy is a BM motion vector search strategy, determining the one or more
refined motion vectors
can include determining the first refined motion vector by searching a first
reference picture around
the first motion vector. The first reference picture can be searched around
the first motion vector
based on the selected motion vector search strategy. The second refined motion
vector can be
determined by searching a second reference picture around the second motion
vector based on the
selected motion vector search strategy. The selected motion vector search
strategy can include a
motion vector difference constraint (e.g., a mirroring constraint wherein the
first and second
motion vector difference have an equal magnitude but opposite sign, a
constraint setting the first
motion vector difference equal to zero, a constraint setting the second motion
vector difference
equal to zero, etc.) In some examples, the first refined motion vector and the
second refined motion
vector can be determined by minimizing a difference between a first reference
block associated
with the first refined motion vector and a second reference block associated
with the second refined
motion vector.,
[0249] At block 1110, the process 1100 includes processing the merge mode
candidate using the
one or more refined motion vectors. For example, the decoding device 112
illustrated in FIG. 1
can process the merge mode candidate using the one or more refined motion
vectors. In some
examples, the decoder engine 116 of decoding device 112 illustrated in FIG. 1
can process the
merge mode candidate using the one or more refined motion vectors.
[0250] In some implementations, the processes (or methods) described herein
can be performed
by a computing device or an apparatus, such as the system 100 shown in FIG. 1.
For example, the
processes can be performed by the encoding device 104 shown in FIG. 1 and FIG.
12, by another
video source-side device or video transmission device, by the decoding device
112 shown in FIG.
1 and FIG. 13, and/or by another client-side device, such as a player device,
a display, or any other
client-side device. In some cases, the computing device or apparatus may
include one or more
input devices, one or more output devices, one or more processors, one or more
microprocessors,
one or more microcomputers, and/or other component(s) that is/are configured
to carry out the
steps of process 1100.
71
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0251] In some examples, the computing device may include a wireless
communication device
such as a mobile device, a tablet computer, an extended reality (XR) device
(e.g., a virtual reality
(VR) device such as a head-mounted display (HMD), an augmented reality (AR)
device such as
an HMD or AR glasses, a mixed reality (MR) device such as an HMD or MR
glasses, etc.), a
desktop computer, a server computer and/or server system, a vehicle or
computing system or
component of a vehicle, or other type of computing device. The components of
the computing
device (e.g., the one or more input devices, one or more output devices, one
or more processors,
one or more microprocessors, one or more microcomputers, and/or other
component) can be
implemented in circuitry. For example, the components can include and/or can
be implemented
using electronic circuits or other electronic hardware, which can include one
or more
programmable electronic circuits (e.g., microprocessors, graphics processing
units (GPUs), digital
signal processors (DSPs), central processing units (CPUs), and/or other
suitable electronic
circuits), and/or can include and/or be implemented using computer software,
firmware, or any
combination thereof, to perform the various operations described herein. In
some examples, the
computing device or apparatus may include a camera configured to capture video
data (e.g., a
video sequence) including video frames. In some examples, a camera or other
capture device that
captures the video data is separate from the computing device, in which case
the computing device
receives or obtains the captured video data. The computing device may include
a network interface
configured to communicate the video data. The network interface may be
configured to
communicate Internet Protocol (IP) based data or other type of data. In some
examples, the
computing device or apparatus may include a display for displaying output
video content, such as
samples of pictures of a video bitstream.
[0252] The processes can be described with respect to logical flow diagrams,
the operation of
which represent a sequence of operations that can be implemented in hardware,
computer
instructions, or a combination thereof. In the context of computer
instructions, the operations
represent computer-executable instructions stored on one or more computer-
readable storage
media that, when executed by one or more processors, perform the recited
operations. Generally,
computer-executable instructions include routines, programs, objects,
components, data structures,
and the like that perform particular functions or implement particular data
types. The order in
which the operations are described is not intended to be construed as a
limitation, and any number
72
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
of the described operations can be combined in any order and/or in parallel to
implement the
processes.
[0253] Additionally, the processes may be performed under the control of one
or more computer
systems configured with executable instructions and may be implemented as code
(e.g., executable
instructions, one or more computer programs, or one or more applications)
executing collectively
on one or more processors, by hardware, or combinations thereof. As noted
above, the code may
be stored on a computer-readable or machine-readable storage medium, for
example, in the form
of a computer program comprising a plurality of instructions executable by one
or more processors.
The computer-readable or machine-readable storage medium may be non-
transitory.
[0254] The coding techniques discussed herein may be implemented in an example
video
encoding and decoding system (e.g., system 100). In some examples, a system
includes a source
device that provides encoded video data to be decoded at a later time by a
destination device. In
particular, the source device provides the video data to destination device
via a computer-readable
medium. The source device and the destination device may comprise any of a
wide range of
devices, including desktop computers, notebook (i.e., laptop) computers,
tablet computers, set-top
boxes, telephone handsets such as so-called "smart" phones, so-called "smart"
pads, televisions,
cameras, display devices, digital media players, video gaming consoles, video
streaming device,
or the like. In some cases, the source device and the destination device may
be equipped for
wireless communication.
[0255] The destination device may receive the encoded video data to be decoded
via the
computer-readable medium. The computer-readable medium may comprise any type
of medium
or device capable of moving the encoded video data from source device to
destination device. In
one example, computer-readable medium may comprise a communication medium to
enable
source device to transmit encoded video data directly to destination device in
real-time. The
encoded video data may be modulated according to a communication standard,
such as a wireless
communication protocol, and transmitted to destination device. The
communication medium may
comprise any wireless or wired communication medium, such as a radio frequency
(RF) spectrum
or one or more physical transmission lines. The communication medium may form
part of a packet-
based network, such as a local area network, a wide-area network, or a global
network such as the
73
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
Internet. The communication medium may include routers, switches, base
stations, or any other
equipment that may be useful to facilitate communication from source device to
destination device.
[0256] In some examples, encoded data may be output from output interface to a
storage device.
Similarly, encoded data may be accessed from the storage device by input
interface. The storage
device may include any of a variety of distributed or locally accessed data
storage media such as
a hard drive, Blu-ray discs, DVDs, CD-ROMs, flash memory, volatile or non-
volatile memory, or
any other suitable digital storage media for storing encoded video data. In a
further example, the
storage device may correspond to a file server or another intermediate storage
device that may
store the encoded video generated by source device. Destination device may
access stored video
data from the storage device via streaming or download. The file server may be
any type of server
capable of storing encoded video data and transmitting that encoded video data
to the destination
device. Example file servers include a web server (e.g., for a website), an
FTP server, network
attached storage (NAS) devices, or a local disk drive. Destination device may
access the encoded
video data through any standard data connection, including an Internet
connection. This may
include a wireless channel (e.g., a Wi-Fi connection), a wired connection
(e.g., DSL, cable modem,
etc.), or a combination of both that is suitable for accessing encoded video
data stored on a file
server. The transmission of encoded video data from the storage device may be
a streaming
transmission, a download transmission, or a combination thereof
[0257] The techniques of this disclosure are not necessarily limited to
wireless applications or
settings. The techniques may be applied to video coding in support of any of a
variety of
multimedia applications, such as over-the-air television broadcasts, cable
television transmissions,
satellite television transmissions, Internet streaming video transmissions,
such as dynamic
adaptive streaming over HTTP (DASH), digital video that is encoded onto a data
storage medium,
decoding of digital video stored on a data storage medium, or other
applications. In some examples,
system may be configured to support one-way or two-way video transmission to
support
applications such as video streaming, video playback, video broadcasting,
and/or video telephony.
[0258] In one example the source device includes a video source, a video
encoder, and a output
interface. The destination device may include an input interface, a video
decoder, and a display
device. The video encoder of source device may be configured to apply the
techniques disclosed
74
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
herein. In other examples, a source device and a destination device may
include other components
or arrangements. For example, the source device may receive video data from an
external video
source, such as an external camera. Likewise, the destination device may
interface with an external
display device, rather than including an integrated display device.
[0259] The example system above is merely one example. Techniques for
processing video data
in parallel may be performed by any digital video encoding and/or decoding
device. Although
generally the techniques of this disclosure are performed by a video encoding
device, the
techniques may also be performed by a video encoder/decoder, typically
referred to as a
"CODEC." Moreover, the techniques of this disclosure may also be performed by
a video
preprocessor. Source device and destination device are merely examples of such
coding devices
in which source device generates coded video data for transmission to
destination device. In some
examples, the source and destination devices may operate in a substantially
symmetrical manner
such that each of the devices include video encoding and decoding components.
Hence, example
systems may support one-way or two-way video transmission between video
devices, e.g., for
video streaming, video playback, video broadcasting, or video telephony.
[0260] The video source may include a video capture device, such as a video
camera, a video
archive containing previously captured video, and/or a video feed interface to
receive video from
a video content provider. As a further alternative, the video source may
generate computer
graphics-based data as the source video, or a combination of live video,
archived video, and
computer-generated video. In some cases, if video source is a video camera,
source device and
destination device may form so-called camera phones or video phones. As
mentioned above,
however, the techniques described in this disclosure may be applicable to
video coding in general,
and may be applied to wireless and/or wired applications. In each case, the
captured, pre-captured,
or computer-generated video may be encoded by the video encoder. The encoded
video
information may then be output by output interface onto the computer-readable
medium.
[0261] As noted the computer-readable medium may include transient media, such
as a wireless
broadcast or wired network transmission, or storage media (that is, non-
transitory storage media),
such as a hard disk, flash drive, compact disc, digital video disc, Blu-ray
disc, or other computer-
readable media. In some examples, a network server (not shown) may receive
encoded video data
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
from the source device and provide the encoded video data to the destination
device, e.g., via
network transmission. Similarly, a computing device of a medium production
facility, such as a
disc stamping facility, may receive encoded video data from the source device
and produce a disc
containing the encoded video data. Therefore, the computer-readable medium may
be understood
to include one or more computer-readable media of various forms, in various
examples.
[0262] The input interface of the destination device receives information from
the computer-
readable medium. The information of the computer-readable medium may include
syntax
information defined by the video encoder, which is also used by the video
decoder, that includes
syntax elements that describe characteristics and/or processing of blocks and
other coded units,
e.g., group of pictures (GOP). A display device displays the decoded video
data to a user, and may
comprise any of a variety of display devices such as a cathode ray tube (CRT),
a liquid crystal
display (LCD), a plasma display, an organic light emitting diode (OLED)
display, or another type
of display device. Various aspects of the application have been described.
[0263] Specific details of the encoding device 104 and the decoding device 112
are shown in
FIG. 12 and FIG. 13, respectively. FIG. 12 is a block diagram illustrating an
example encoding
device 104 that may implement one or more of the techniques described in this
disclosure.
Encoding device 104 may, for example, generate the syntax structures described
herein (e.g., the
syntax structures of a VPS, SPS, PPS, or other syntax elements). Encoding
device 104 may
perform intra-prediction and inter-prediction coding of video blocks within
video slices. As
previously described, intra-coding relies, at least in part, on spatial
prediction to reduce or remove
spatial redundancy within a given video frame or picture. Inter-coding relies,
at least in part, on
temporal prediction to reduce or remove temporal redundancy within adjacent or
surrounding
frames of a video sequence. Intra-mode (I mode) may refer to any of several
spatial based
compression modes. Inter-modes, such as uni-directional prediction (P mode) or
bi-prediction (B
mode), may refer to any of several temporal-based compression modes.
[0264] The encoding device 104 includes a partitioning unit 35, prediction
processing unit 41,
filter unit 63, picture memory 64, summer 50, transform processing unit 52,
quantization unit 54,
and entropy encoding unit 56. Prediction processing unit 41 includes motion
estimation unit 42,
motion compensation unit 44, and intra-prediction processing unit 46. For
video block
76
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
reconstruction, encoding device 104 also includes inverse quantization unit
58, inverse transform
processing unit 60, and summer 62. Filter unit 63 is intended to represent one
or more loop filters
such as a deblocking filter, an adaptive loop filter (ALF), and a sample
adaptive offset (SAO) filter.
Although filter unit 63 is shown in FIG. 12 as being an in loop filter, in
other configurations, filter
unit 63 may be implemented as a post loop filter. A post processing device 57
may perform
additional processing on encoded video data generated by the encoding device
104. The techniques
of this disclosure may in some instances be implemented by the encoding device
104. In other
instances, however, one or more of the techniques of this disclosure may be
implemented by post
processing device 57.
[0265] As shown in FIG. 12, the encoding device 104 receives video data, and
partitioning unit
35 partitions the data into video blocks. The partitioning may also include
partitioning into slices,
slice segments, tiles, or other larger units, as wells as video block
partitioning, e.g., according to a
quadtree structure of LCUs and CUs. The encoding device 104 generally
illustrates the
components that encode video blocks within a video slice to be encoded. The
slice may be divided
into multiple video blocks (and possibly into sets of video blocks referred to
as tiles). Prediction
processing unit 41 may select one of a plurality of possible coding modes,
such as one of a plurality
of intra-prediction coding modes or one of a plurality of inter-prediction
coding modes, for the
current video block based on error results (e.g., coding rate and the level of
distortion, or the like).
Prediction processing unit 41 may provide the resulting intra- or inter-coded
block to summer 50
to generate residual block data and to summer 62 to reconstruct the encoded
block for use as a
reference picture.
[0266] Intra-prediction processing unit 46 within prediction processing unit
41 may perform
intra-prediction coding of the current video block relative to one or more
neighboring blocks in
the same frame or slice as the current block to be coded to provide spatial
compression. Motion
estimation unit 42 and motion compensation unit 44 within prediction
processing unit 41 perform
inter-predictive coding of the current video block relative to one or more
predictive blocks in one
or more reference pictures to provide temporal compression.
[0267] Motion estimation unit 42 may be configured to determine the inter-
prediction mode for
a video slice according to a predetermined pattern for a video sequence. The
predetermined pattern
77
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
may designate video slices in the sequence as P slices, B slices, or GPB
slices. Motion estimation
unit 42 and motion compensation unit 44 may be highly integrated, but are
illustrated separately
for conceptual purposes. Motion estimation, performed by motion estimation
unit 42, is the process
of generating motion vectors, which estimate motion for video blocks. A motion
vector, for
example, may indicate the displacement of a prediction unit (PU) of a video
block within a current
video frame or picture relative to a predictive block within a reference
picture.
[0268] A predictive block is a block that is found to closely match the PU of
the video block to
be coded in terms of pixel difference, which may be determined by sum of
absolute difference
(SAD), sum of square difference (S SD), or other difference metrics. In some
examples, the
encoding device 104 may calculate values for sub-integer pixel positions of
reference pictures
stored in picture memory 64. For example, the encoding device 104 may
interpolate values of one-
quarter pixel positions, one-eighth pixel positions, or other fractional pixel
positions of the
reference picture. Therefore, motion estimation unit 42 may perform a motion
search relative to
the full pixel positions and fractional pixel positions and output a motion
vector with fractional
pixel precision.
[0269] Motion estimation unit 42 calculates a motion vector for a PU of a
video block in an
inter-coded slice by comparing the position of the PU to the position of a
predictive block of a
reference picture. The reference picture may be selected from a first
reference picture list (List 0)
or a second reference picture list (List 1), each of which identify one or
more reference pictures
stored in picture memory 64. Motion estimation unit 42 sends the calculated
motion vector to
entropy encoding unit 56 and motion compensation unit 44.
[0270] Motion compensation, performed by motion compensation unit 44, may
involve fetching
or generating the predictive block based on the motion vector determined by
motion estimation,
possibly performing interpolations to sub-pixel precision. Upon receiving the
motion vector for
the PU of the current video block, motion compensation unit 44 may locate the
predictive block to
which the motion vector points in a reference picture list. The encoding
device 104 forms a residual
video block by subtracting pixel values of the predictive block from the pixel
values of the current
video block being coded, forming pixel difference values. The pixel difference
values form
residual data for the block, and may include both luma and chroma difference
components.
78
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
Summer 50 represents the component or components that perform this subtraction
operation.
Motion compensation unit 44 may also generate syntax elements associated with
the video blocks
and the video slice for use by the decoding device 112 in decoding the video
blocks of the video
slice.
[0271] Intra-prediction processing unit 46 may intra-predict a current block,
as an alternative to
the inter-prediction performed by motion estimation unit 42 and motion
compensation unit 44, as
described above. In particular, intra-prediction processing unit 46 may
determine an intra-
prediction mode to use to encode a current block. In some examples, intra-
prediction processing
unit 46 may encode a current block using various intra-prediction modes, e.g.,
during separate
encoding passes, and intra-prediction processing unit 46 may select an
appropriate intra-prediction
mode to use from the tested modes. For example, intra-prediction processing
unit 46 may calculate
rate-distortion values using a rate-distortion analysis for the various tested
intra-prediction modes,
and may select the intra-prediction mode having the best rate-distortion
characteristics among the
tested modes. Rate-distortion analysis generally determines an amount of
distortion (or error)
between an encoded block and an original, unencoded block that was encoded to
produce the
encoded block, as well as a bit rate (that is, a number of bits) used to
produce the encoded block.
Intra-prediction processing unit 46 may calculate ratios from the distortions
and rates for the
various encoded blocks to determine which intra-prediction mode exhibits the
best rate-distortion
value for the block.
[0272] In any case, after selecting an intra-prediction mode for a block,
intra-prediction
processing unit 46 may provide information indicative of the selected intra-
prediction mode for
the block to entropy encoding unit 56. Entropy encoding unit 56 may encode the
information
indicating the selected intra-prediction mode. The encoding device 104 may
include in the
transmitted bitstream configuration data definitions of encoding contexts for
various blocks as well
as indications of a most probable intra-prediction mode, an intra-prediction
mode index table, and
a modified intra-prediction mode index table to use for each of the contexts.
The bitstream
configuration data may include a plurality of intra-prediction mode index
tables and a plurality of
modified intra-prediction mode index tables (also referred to as codeword
mapping tables).
79
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0273] After prediction processing unit 41 generates the predictive block for
the current video
block via either inter-prediction or intra-prediction, the encoding device 104
forms a residual video
block by subtracting the predictive block from the current video block. The
residual video data in
the residual block may be included in one or more TUs and applied to transform
processing unit
52. Transform processing unit 52 transforms the residual video data into
residual transform
coefficients using a transform, such as a discrete cosine transform (DCT) or a
conceptually similar
transform. Transform processing unit 52 may convert the residual video data
from a pixel domain
to a transform domain, such as a frequency domain.
[0274] Transform processing unit 52 may send the resulting transform
coefficients to
quantization unit 54. Quantization unit 54 quantizes the transform
coefficients to further reduce
bit rate. The quantization process may reduce the bit depth associated with
some or all of the
coefficients. The degree of quantization may be modified by adjusting a
quantization parameter.
In some examples, quantization unit 54 may then perform a scan of the matrix
including the
quantized transform coefficients. Alternatively, entropy encoding unit 56 may
perform the scan.
[0275] Following quantization, entropy encoding unit 56 entropy encodes the
quantized
transform coefficients. For example, entropy encoding unit 56 may perform
context adaptive
variable length coding (CAVLC), context adaptive binary arithmetic coding
(CABAC), syntax-
based context-adaptive binary arithmetic coding (SBAC), probability interval
partitioning entropy
(PIPE) coding or another entropy encoding technique. Following the entropy
encoding by entropy
encoding unit 56, the encoded bitstream may be transmitted to the decoding
device 112, or
archived for later transmission or retrieval by the decoding device 112.
Entropy encoding unit 56
may also entropy encode the motion vectors and the other syntax elements for
the current video
slice being coded.
[0276] Inverse quantization unit 58 and inverse transform processing unit 60
apply inverse
quantization and inverse transformation, respectively, to reconstruct the
residual block in the pixel
domain for later use as a reference block of a reference picture. Motion
compensation unit 44 may
calculate a reference block by adding the residual block to a predictive block
of one of the reference
pictures within a reference picture list. Motion compensation unit 44 may also
apply one or more
interpolation filters to the reconstructed residual block to calculate sub-
integer pixel values for use
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
in motion estimation. Summer 62 adds the reconstructed residual block to the
motion compensated
prediction block produced by motion compensation unit 44 to produce a
reference block for
storage in picture memory 64. The reference block may be used by motion
estimation unit 42 and
motion compensation unit 44 as a reference block to inter-predict a block in a
subsequent video
frame or picture.
[0277] In this manner, the encoding device 104 of FIG. 12 represents an
example of a video
encoder configured to perform any of the techniques described herein,
including the process
described above with respect to FIG. 11. In some cases, some of the techniques
of this disclosure
may also be implemented by post processing device 57.
[0278] FIG. 13 is a block diagram illustrating an example decoding device 112.
The decoding
device 112 includes an entropy decoding unit 80, prediction processing unit
81, inverse
quantization unit 86, inverse transform processing unit 88, summer 90, filter
unit 91, and picture
memory 92. Prediction processing unit 81 includes motion compensation unit 82
and intra
prediction processing unit 84. The decoding device 112 may, in some examples,
perform a
decoding pass generally reciprocal to the encoding pass described with respect
to the encoding
device 104 from FIG. 12.
[0279] During the decoding process, the decoding device 112 receives an
encoded video
bitstream that represents video blocks of an encoded video slice and
associated syntax elements
sent by the encoding device 104. In some aspects, the decoding device 112 may
receive the
encoded video bitstream from the encoding device 104. In some aspects, the
decoding device 112
may receive the encoded video bitstream from a network entity 79, such as a
server, a media-aware
network element (MANE), a video editor/splicer, or other such device
configured to implement
one or more of the techniques described above. Network entity 79 may or may
not include the
encoding device 104. Some of the techniques described in this disclosure may
be implemented by
network entity 79 prior to network entity 79 transmitting the encoded video
bitstream to the
decoding device 112. In some video decoding systems, network entity 79 and the
decoding device
112 may be parts of separate devices, while in other instances, the
functionality described with
respect to network entity 79 may be performed by the same device that
comprises the decoding
device 112.
81
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0280] The entropy decoding unit 80 of the decoding device 112 entropy decodes
the bitstream
to generate quantized coefficients, motion vectors, and other syntax elements.
Entropy decoding
unit 80 forwards the motion vectors and other syntax elements to prediction
processing unit 81.
The decoding device 112 may receive the syntax elements at the video slice
level and/or the video
block level. Entropy decoding unit 80 may process and parse both fixed-length
syntax elements
and variable-length syntax elements in or more parameter sets, such as a VPS,
SPS, and PPS.
[0281] When the video slice is coded as an intra-coded (I) slice, intra
prediction processing unit
84 of prediction processing unit 81 may generate prediction data for a video
block of the current
video slice based on a signaled intra-prediction mode and data from previously
decoded blocks of
the current frame or picture. When the video frame is coded as an inter-coded
(i.e., B, P or GPB)
slice, motion compensation unit 82 of prediction processing unit 81 produces
predictive blocks for
a video block of the current video slice based on the motion vectors and other
syntax elements
received from entropy decoding unit 80. The predictive blocks may be produced
from one of the
reference pictures within a reference picture list. The decoding device 112
may construct the
reference frame lists, List 0 and List 1, using default construction
techniques based on reference
pictures stored in picture memory 92.
[0282] Motion compensation unit 82 determines prediction information for a
video block of the
current video slice by parsing the motion vectors and other syntax elements,
and uses the prediction
information to produce the predictive blocks for the current video block being
decoded. For
example, motion compensation unit 82 may use one or more syntax elements in a
parameter set to
determine a prediction mode (e.g., intra- or inter-prediction) used to code
the video blocks of the
video slice, an inter-prediction slice type (e.g., B slice, P slice, or GPB
slice), construction
information for one or more reference picture lists for the slice, motion
vectors for each inter-
encoded video block of the slice, inter-prediction status for each inter-coded
video block of the
slice, and other information to decode the video blocks in the current video
slice.
[0283] Motion compensation unit 82 may also perform interpolation based on
interpolation
filters. Motion compensation unit 82 may use interpolation filters as used by
the encoding device
104 during encoding of the video blocks to calculate interpolated values for
sub-integer pixels of
reference blocks. In this case, motion compensation unit 82 may determine the
interpolation filters
82
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
used by the encoding device 104 from the received syntax elements, and may use
the interpolation
filters to produce predictive blocks.
[0284] Inverse quantization unit 86 inverse quantizes, or de-quantizes, the
quantized transform
coefficients provided in the bitstream and decoded by entropy decoding unit
80. The inverse
quantization process may include use of a quantization parameter calculated by
the encoding
device 104 for each video block in the video slice to determine a degree of
quantization and,
likewise, a degree of inverse quantization that should be applied. Inverse
transform processing unit
88 applies an inverse transform (e.g., an inverse DCT or other suitable
inverse transform), an
inverse integer transform, or a conceptually similar inverse transform
process, to the transform
coefficients in order to produce residual blocks in the pixel domain.
[0285] After motion compensation unit 82 generates the predictive block for
the current video
block based on the motion vectors and other syntax elements, the decoding
device 112 forms a
decoded video block by summing the residual blocks from inverse transform
processing unit 88
with the corresponding predictive blocks generated by motion compensation unit
82. Summer 90
represents the component or components that perform this summation operation.
If desired, loop
filters (either in the coding loop or after the coding loop) may also be used
to smooth pixel
transitions, or to otherwise improve the video quality. Filter unit 91 is
intended to represent one or
more loop filters such as a deblocking filter, an adaptive loop filter (ALF),
and a sample adaptive
offset (SAO) filter. Although filter unit 91 is shown in FIG. 8 as being an in
loop filter, in other
configurations, filter unit 91 may be implemented as a post loop filter. The
decoded video blocks
in a given frame or picture are then stored in picture memory 92, which stores
reference pictures
used for subsequent motion compensation. Picture memory 92 also stores decoded
video for later
presentation on a display device, such as video destination device 122 shown
in FIG. 1.
[0286] In this manner, the decoding device 112 of FIG. 13 represents an
example of a video
decoder configured to perform any of the techniques described herein,
including the process
described above with respect to FIG. 11.
[0287] As used herein, the term "computer-readable medium" includes, but is
not limited to,
portable or non-portable storage devices, optical storage devices, and various
other mediums
capable of storing, containing, or carrying instruction(s) and/or data. A
computer-readable medium
83
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
may include a non-transitory medium in which data can be stored and that does
not include carrier
waves and/or transitory electronic signals propagating wirelessly or over
wired connections.
Examples of a non-transitory medium may include, but are not limited to, a
magnetic disk or tape,
optical storage media such as compact disk (CD) or digital versatile disk
(DVD), flash memory,
memory or memory devices. A computer-readable medium may have stored thereon
code and/or
machine-executable instructions that may represent a procedure, a function, a
subprogram, a
program, a routine, a subroutine, a module, a software package, a class, or
any combination of
instructions, data structures, or program statements. A code segment may be
coupled to another
code segment or a hardware circuit by passing and/or receiving information,
data, arguments,
parameters, or memory contents. Information, arguments, parameters, data, etc.
may be passed,
forwarded, or transmitted via any suitable means including memory sharing,
message passing,
token passing, network transmission, or the like.
[0288] In some aspects the computer-readable storage devices, mediums, and
memories can
include a cable or wireless signal containing a bit stream and the like.
However, when mentioned,
non-transitory computer-readable storage media expressly exclude media such as
energy, carrier
signals, electromagnetic waves, and signals per se.
[0289] Specific details are provided in the description above to provide a
thorough
understanding of the aspects and examples provided herein. However, it will be
understood by one
of ordinary skill in the art that the aspects may be practiced without these
specific details. For
clarity of explanation, in some instances the present technology may be
presented as including
individual functional blocks including functional blocks comprising devices,
device components,
steps or routines in a method embodied in software, or combinations of
hardware and software.
Additional components may be used other than those shown in the figures and/or
described herein.
For example, circuits, systems, networks, processes, and other components may
be shown as
components in block diagram form in order not to obscure the aspects in
unnecessary detail. In
other instances, well-known circuits, processes, algorithms, structures, and
techniques may be
shown without unnecessary detail in order to avoid obscuring the aspects.
[0290] Individual aspects may be described above as a process or method which
is depicted as
a flowchart, a flow diagram, a data flow diagram, a structure diagram, or a
block diagram.
84
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
Although a flowchart may describe the operations as a sequential process, many
of the operations
can be performed in parallel or concurrently. In addition, the order of the
operations may be re-
arranged. A process is terminated when its operations are completed, but could
have additional
steps not included in a figure. A process may correspond to a method, a
function, a procedure, a
subroutine, a subprogram, etc. When a process corresponds to a function, its
termination can
correspond to a return of the function to the calling function or the main
function.
[0291] Processes and methods according to the above-described examples can be
implemented
using computer-executable instructions that are stored or otherwise available
from computer-
readable media. Such instructions can include, for example, instructions and
data which cause or
otherwise configure a general purpose computer, special purpose computer, or a
processing device
to perform a certain function or group of functions. Portions of computer
resources used can be
accessible over a network. The computer executable instructions may be, for
example, binaries,
intermediate format instructions such as assembly language, firmware, source
code, etc. Examples
of computer-readable media that may be used to store instructions, information
used, and/or
information created during methods according to described examples include
magnetic or optical
disks, flash memory, USB devices provided with non-volatile memory, networked
storage devices,
and so on.
[0292] Devices implementing processes and methods according to these
disclosures can include
hardware, software, firmware, middleware, microcode, hardware description
languages, or any
combination thereof, and can take any of a variety of form factors. When
implemented in software,
firmware, middleware, or microcode, the program code or code segments to
perform the necessary
tasks (e.g., a computer-program product) may be stored in a computer-readable
or machine-
readable medium. A processor(s) may perform the necessary tasks. Typical
examples of form
factors include laptops, smart phones, mobile phones, tablet devices or other
small form factor
personal computers, personal digital assistants, rackmount devices, standalone
devices, and so on.
Functionality described herein also can be embodied in peripherals or add-in
cards. Such
functionality can also be implemented on a circuit board among different chips
or different
processes executing in a single device, by way of further example.
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0293] The instructions, media for conveying such instructions, computing
resources for
executing them, and other structures for supporting such computing resources
are example means
for providing the functions described in the disclosure.
[0294] In the foregoing description, aspects of the application are described
with reference to
specific aspects thereof, but those skilled in the art will recognize that the
application is not limited
thereto. Thus, while illustrative aspects of the application have been
described in detail herein, it
is to be understood that the inventive concepts may be otherwise variously
embodied and
employed, and that the appended claims are intended to be construed to include
such variations,
except as limited by the prior art. Various features and aspects of the above-
described application
may be used individually or jointly. Further, aspects can be utilized in any
number of environments
and applications beyond those described herein without departing from the
broader spirit and scope
of the specification. The specification and drawings are, accordingly, to be
regarded as illustrative
rather than restrictive. For the purposes of illustration, methods were
described in a particular
order. It should be appreciated that in alternate aspects, the methods may be
performed in a
different order than that described.
[0295] One of ordinary skill will appreciate that the less than ("<") and
greater than (">")
symbols or terminology used herein can be replaced with less than or equal to
(" ") and greater
than or equal to (" ") symbols, respectively, without departing from the scope
of this description.
[0296] Where components are described as being "configured to" perform certain
operations,
such configuration can be accomplished, for example, by designing electronic
circuits or other
hardware to perform the operation, by programming programmable electronic
circuits (e.g.,
microprocessors, or other suitable electronic circuits) to perform the
operation, or any combination
thereof.
[0297] The phrase "coupled to" refers to any component that is physically
connected to another
component either directly or indirectly, and/or any component that is in
communication with
another component (e.g., connected to the other component over a wired or
wireless connection,
and/or other suitable communication interface) either directly or indirectly.
86
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0298] Claim language or other language in the disclosure reciting "at least
one of' a set and/or
"one or more" of a set indicates that one member of the set or multiple
members of the set (in any
combination) satisfy the claim. For example, claim language reciting "at least
one of A and B" and
"at least one of A or B" means A, B, or A and B. In another example, claim
language reciting "at
least one of A, B, and C" and "at least one of A, B, or C" means A, B, C, or A
and B, or A and C,
or B and C, or A and B and C. The language "at least one of' a set and/or "one
or more" of a set
does not limit the set to the items listed in the set. For example, claim
language reciting "at least
one of A and B" and "at least one of A or B" can mean A, B, or A and B, and
can additionally
include items not listed in the set of A and B.
[0299] The various illustrative logical blocks, modules, circuits, and
algorithm steps described
in connection with the aspects disclosed herein may be implemented as
electronic hardware,
computer software, firmware, or combinations thereof. To clearly illustrate
this interchangeability
of hardware and software, various illustrative components, blocks, modules,
circuits, and steps
have been described above generally in terms of their functionality. Whether
such functionality is
implemented as hardware or software depends upon the particular application
and design
constraints imposed on the overall system. Skilled artisans may implement the
described
functionality in varying ways for each particular application, but such
implementation decisions
should not be interpreted as causing a departure from the scope of the present
application.
[0300] The techniques described herein may also be implemented in electronic
hardware,
computer software, firmware, or any combination thereof. Such techniques may
be implemented
in any of a variety of devices such as general purposes computers, wireless
communication device
handsets, or integrated circuit devices having multiple uses including
application in wireless
communication device handsets and other devices. Any features described as
modules or
components may be implemented together in an integrated logic device or
separately as discrete
but interoperable logic devices. If implemented in software, the techniques
may be realized at least
in part by a computer-readable data storage medium comprising program code
including
instructions that, when executed, performs one or more of the methods
described above. The
computer-readable data storage medium may form part of a computer program
product, which
may include packaging materials. The computer-readable medium may comprise
memory or data
storage media, such as random access memory (RAM) such as synchronous dynamic
random
87
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
access memory (SDRAM), read-only memory (ROM), non-volatile random access
memory
(NVRAM), electrically erasable programmable read-only memory (EEPROM), FLASH
memory,
magnetic or optical data storage media, and the like. The techniques
additionally, or alternatively,
may be realized at least in part by a computer-readable communication medium
that carries or
communicates program code in the form of instructions or data structures and
that can be accessed,
read, and/or executed by a computer, such as propagated signals or waves.
[0301] The program code may be executed by a processor, which may include one
or more
processors, such as one or more digital signal processors (DSPs), general
purpose microprocessors,
an application specific integrated circuits (ASICs), field programmable logic
arrays (FPGAs), or
other equivalent integrated or discrete logic circuitry. Such a processor may
be configured to
perform any of the techniques described in this disclosure. A general purpose
processor may be a
microprocessor; but in the alternative, the processor may be any conventional
processor, controller,
microcontroller, or state machine. A processor may also be implemented as a
combination of
computing devices, e.g., a combination of a DSP and a microprocessor, a
plurality of
microprocessors, one or more microprocessors in conjunction with a DSP core,
or any other such
configuration. Accordingly, the term "processor," as used herein may refer to
any of the foregoing
structure, any combination of the foregoing structure, or any other structure
or apparatus suitable
for implementation of the techniques described herein. In addition, in some
aspects, the
functionality described herein may be provided within dedicated software
modules or hardware
modules configured for encoding and decoding, or incorporated in a combined
video encoder-
decoder (CODEC).
[0302] Illustrative aspects of the present disclosure include:
[0303] Aspect 1: An apparatus for processing video data, comprising: memory;
and one or more
processors coupled to the memory. The one or more processors are configured
to: obtain a current
picture of the video data; obtain reference pictures for the current picture
from the video data;
determine a merge mode candidate from the current picture; identify a first
motion vector and a
second motion vector for the merge mode candidate; select a motion vector
search strategy for the
merge mode candidate from a plurality of motion vector search strategies;
derive refined motion
vectors from the first motion vector, the second motion vector, and the
reference pictures using
88
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
the motion vector search strategy; and process the merge mode candidate using
the refined motion
vectors.
[0304] Aspect 2: The apparatus of Aspect 1, wherein the merge mode candidate
is selected from
a merge candidate list.
[0305] Aspect 3: The apparatus of Aspect 2, wherein the merge candidate list
is constructed from
one or more of a spatial motion vector predictor from spatial neighbor blocks
of the merge mode
candidate, a temporal motion vector predictor from co-located blocks of the
merge mode
candidate, a history based motion vector predictor from a history table, a
pairwise average motion
vector predictor, and a zero value motion vector.
[0306] Aspect 4: The apparatus of any of Aspects 1 to 3, wherein the one or
more processors
being configured to generate a motion vector bi-prediction signal using the
first motion vector and
the second motion vector by averaging two prediction signals obtained from two
different
reference pictures.
[0307] Aspect 5: The apparatus of any of Aspects 1 to 4, wherein the plurality
of motion vector
search strategies includes a fractional sample refinement strategy.
[0308] Aspect 6: The apparatus of Aspect 5, wherein the plurality of motion
vector search
strategies includes a bi-directional optical flow strategy.
[0309] Aspect 7: The apparatus of Aspect 6, wherein the plurality of motion
vector search
strategies includes a sub-block based bilateral matching motion vector
refinement strategy.
[0310] Aspect 8: The apparatus of any of Aspects 1 to 7, wherein the first
motion vector and the
second motion vector are associated with one or more constraints.
[0311] Aspect 9: The apparatus of Aspect 8, wherein the one or more
constraints includes a
mirroring constraint.
[0312] Aspect 10: The apparatus of any of Aspects 1 to 9, wherein the one or
more constraints
includes a zero value constraint for the first motion vector difference.
89
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0313] Aspect 11: The apparatus of any of Aspects 1 to 9, wherein the one or
more constraints
includes a zero value constraint for the second motion vector difference.
[0314] Aspect 12: The apparatus of any of Aspects 1 to 11, wherein the video
data includes
syntax indicating the one or more constraints.
[0315] Aspect 13: The apparatus of any of Aspects 1 to 12, wherein the motion
vector search
strategy comprises a multi-pass decoder side motion vector refinement
strategy.
[0316] Aspect 14: The apparatus of Aspect 13, wherein the multi-pass decoder
side motion
vector refinement strategy includes two or more refinement passes of a same
refinement type.
[0317] Aspect 15: The apparatus of Aspect 14, wherein the multi-pass decoder
side motion
vector refinement strategy includes one or more refinement passes of a type
different than the same
refinement type.
[0318] Aspect 16: The apparatus of any of Aspects 14 or 15, wherein the two or
more refinement
passes of the same refinement type are a block based bilateral matching motion
vector refinement,
a sub-block based bilateral matching motion vector refinement, or a sub-block
based bi-directional
optical flow motion vector refinement.
[0319] Aspect 17: The apparatus of any of Aspects 1 to 16, wherein the
plurality of motion
vector search strategies comprise multiple subsets of multi-pass strategies.
[0320] Aspect 18: The apparatus of Aspect 17, wherein the multiple subsets of
multi-pass
strategies are signaled in one or more syntax elements of the video data.
[0321] Aspect 19: The apparatus of any of Aspects 1 to 18, wherein deriving
the refined motion
vectors comprises calculating matching costs for a plurality of candidate
motion vector pairs
according to the motion vector search strategy.
[0322] Aspect 20: The apparatus of any of Aspects 1 to 19, wherein the motion
vector search
strategy is selected adaptively based on matching costs determined from the
video data.
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0323] Aspect 21: The apparatus of any of Aspects 1 to 19, wherein the motion
vector search
strategy is selected to adaptively set a number of passes of the motion vector
search strategy based
on the video data.
[0324] Aspect 22: The apparatus of any of Aspects 1 to 19, wherein the motion
vector search
strategy is selected to adaptively set a search pattern for determining
candidates for the refined
motion vectors based on the video data.
[0325] Aspect 23: The apparatus of any of Aspects 1 to 19, wherein the motion
vector search
strategy is selected to adaptively set a set of criteria for generating a list
of candidates for the
refined motion vectors based on the video data.
[0326] Aspect 24: The apparatus of any of Aspects 1 to 19, wherein the motion
vector search
strategy is adaptively performed based on decoder side motion vector
refinement constraints from
the video data.
[0327] Aspect 25: The apparatus of any of Aspects 1 to 19, wherein the motion
vector search
strategy is adaptively performed based on a block size for the merge mode
candidate in the video
data.
[0328] Aspect 26: The apparatus of any of Aspects 1 to 25, the one or more
processors being
configured to disable multiple hypothesis prediction.
[0329] Aspect 27: The apparatus of any of Aspects 1 to 26, the one or more
processors being
configured to perform multiple hypothesis prediction in conjunction with the
motion vector search
strategy.
[0330] Aspect 28: The apparatus of any of Aspects 1 to 27, the one or more
processors being
configured to: generate a merge candidate list including the merge mode
candidate.
[0331] Aspect 29: The apparatus of Aspect 28, wherein, to generate the merge
candidate list, the
one or more processors are configured to: determine, based on one or more
conditions associated
with an adaptive merge mode (e.g., conditions for the adaptive bm mode), one
or more default
candidates for adding to the merge candidate list based on a number of
candidates in the merge
candidate list being less than a maximum number of candidates.
91
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0332] Aspect 30: The apparatus of Aspect 28, wherein, to generate the merge
candidate list, the
one or more processors are configured to: determine, based on one or more
conditions associated
with a constraint associated with an adaptive merge mode (e.g., conditions
according to the
bm dir), one or more default candidates for adding to the merge candidate list
based on a number
of candidates in the merge candidate list being less than a maximum number of
candidates.
[0333] Aspect 31: The apparatus of any of Aspects 1 to 30, wherein the
apparatus is a mobile
device.
[0334] Aspect 32: The apparatus of any of Aspects 1 to 31, further comprising
a camera
configured to capture one or more frames.
[0335] Aspect 33: The apparatus of any of Aspects 1 to 32, further comprising
a display
configured to display one or more frames.
[0336] Aspect 34: A method of processing video data in accordance with any of
the operations
of Aspects 1 to 33.
[0337] Aspect 35: A computer readable storage medium comprising instructions
that, when
executed by one or more processors of a device, cause the device to perform
the operations of any
of Aspects 1 to 33.
[0338] Aspect 36: An apparatus comprising one or more means for performing any
of the
operations of Aspects 1 to 33.
[0339] Aspect 37: An apparatus for processing video data, comprising: at least
one memory; and
at least one processor coupled to the at least one memory, the at least one
processor configured to:
obtain one or more reference pictures for a current picture; identify a first
motion vector and a
second motion vector for a merge mode candidate; determine a selected motion
vector search
strategy for the merge mode candidate from a plurality of motion vector search
strategies;
determine, using the selected motion vector search strategy, one or more
refined motion vectors
based on at least one of the first motion vector or the second motion vector
and the one or more
reference pictures; and process the merge mode candidate using the one or more
refined motion
vectors.
92
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0340] Aspect 38. The apparatus of Aspect 37, wherein the selected motion
vector search
strategy is associated with one or more constraints based at least on at least
one of the first motion
vector or the second motion vector.
[0341] Aspect 39. The apparatus of Aspect 38, wherein the one or more
constraints are
determined for a block of the video data based on a syntax element signaled
for the block.
[0342] Aspect 40. The apparatus of any of Aspects 38 or 39, wherein the one or
more constraints
are associated with at least one of a first motion vector difference
associated with the first motion
vector or a second motion vector difference associated with the second motion
vector.
[0343] Aspect 41. The apparatus of Aspect 40, wherein the one or more refined
motion vectors
include a first refined motion vector and a second refined motion vector, and
wherein the at least
one processor is configured to: determine the first motion vector difference
as a difference between
the first refined motion vector and the first motion vector; and determine the
second motion vector
difference as a difference between the second refined motion vector and the
second motion vector.
[0344] Aspect 42. The apparatus of any of Aspects 40 or 41, wherein the one or
more constraints
include a mirroring constraint for the first motion vector difference and the
second motion vector
difference, and wherein the first motion vector difference and the second
motion vector difference
have a same magnitude and a different sign.
[0345] Aspect 43. The apparatus of any of Aspects 40 to 42, wherein the one or
more constraints
include a zero value constraint for at least one of the first motion vector
difference or the second
motion vector difference.
[0346] Aspect 44. The apparatus of Aspect 43, wherein, based on the zero value
constraint, the
at least one processor is configured to determine the one or more refined
motion vectors using the
selected motion vector search strategy by maintaining a first one of the first
motion vector
difference or the second motion vector difference as a fixed value and
searching relative to a
second one of the first motion vector difference or the second motion vector
difference.
[0347] Aspect 45. The apparatus of any of Aspects 37 to 44, wherein the
selected motion vector
search strategy is a bilateral matching (BM) motion vector search strategy.
93
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0348] Aspect 46. The apparatus of any of Aspects 37 to 45, wherein the at
least one processor
is configured to determine the one or more refined motion vectors based on one
or more constraints
associated with the selected motion vector search strategy, and wherein, to
determine the one or
more refined motion vectors based on the one or more constraints, the at least
one processor is
configured to: determine a first refined motion vector by searching a first
reference picture around
the first motion vector based on the selected motion vector search strategy;
and determine a second
refined motion vector by searching a second reference picture around the
second motion vector
based on the selected motion vector search strategy; wherein the one or more
constraints include
a motion vector difference constraint.
[0349] Aspect 47. The apparatus of Aspect 46, wherein, to determine the first
refined motion
vector and the second refined motion vector, the at least one processor is
configured to: minimize
a difference between a first reference block associated with the first refined
motion vector and a
second reference block associated with the second refined motion vector.
[0350] Aspect 48. The apparatus of any of Aspects 37 to 47, wherein the
plurality of motion
vector search strategies includes at least two of a multi-pass decoder side
motion vector refinement
strategy, a fractional sample refinement strategy, a bi-directional optical
flow strategy, or a sub-
block based bilateral matching motion vector refinement strategy.
[0351] Aspect 49. The apparatus of any of Aspects 37 to 48, wherein the
selected motion vector
search strategy comprises a multi-pass decoder side motion vector refinement
strategy.
[0352] Aspect 50. The apparatus of Aspect 49, wherein the multi-pass decoder
side motion
vector refinement strategy includes at least one of one or more block based
bilateral matching
motion vector refinement passes or one or more sub-block based motion vector
refinement passes.
[0353] Aspect 51. The apparatus of Aspect 50, wherein the at least one
processor is configured
to: perform the one or more block based bilateral matching motion vector
refinement passes using
a first constraint associated with at least one of a first motion vector
difference or a second motion
vector difference; and perform the one or more sub-block based motion vector
refinement passes
using a second constraint associated with at least one of the first motion
vector difference or the
second motion vector difference, wherein the first constraint is different
than the second constraint.
94
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0354] Aspect 52. The apparatus of any of Aspects 50 or 51, wherein the one or
more sub-block
based motion vector refinement passes include at least one of a sub-block
based bilateral matching
motion vector refinement pass or a sub-block based bi-directional optical flow
motion vector
refinement pass.
[0355] Aspect 53. The apparatus of any of Aspects 37 to 52, wherein the
apparatus is a wireless
communication device.
[0356] Aspect 54. The apparatus of any of Aspects 37 to 53, wherein the at
least one processor
is configured to determine the one or more refined motion vectors for a block
of the video data,
and wherein the merge mode candidate includes a neighboring block of the block
[0357] Aspect 55: A method for processing video data, comprising: obtaining
one or more
reference pictures for a current picture; identifying a first motion vector
and a second motion vector
for a merge mode candidate; determining a selected motion vector search
strategy for the merge
mode candidate from a plurality of motion vector search strategies;
determining, using the selected
motion vector search strategy, one or more refined motion vectors based on at
least one of the first
motion vector or the second motion vector and the one or more reference
pictures; and processing
the merge mode candidate using the one or more refined motion vectors.
[0358] Aspect 56. The method of Aspect 55, wherein the selected motion vector
search strategy
is associated with one or more constraints based at least on at least one of
the first motion vector
or the second motion vector.
[0359] Aspect 57. The method of Aspect 56, wherein the one or more constraints
are determined
for a block of the video data based on a syntax element signaled for the
block.
[0360] Aspect 58. The method of any of Aspects 56 or 57, wherein the one or
more constraints
are associated with at least one of a first motion vector difference
associated with the first motion
vector or a second motion vector difference associated with the second motion
vector.
[0361] Aspect 59. The method of Aspect 58, wherein the one or more refined
motion vectors
include a first refined motion vector and a second refined motion vector, the
method further
comprising: determining the first motion vector difference as a difference
between the first refined
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
motion vector and the first motion vector; and determining the second motion
vector difference as
a difference between the second refined motion vector and the second motion
vector.
[0362] Aspect 60. The method of any of Aspects 58 or 59, wherein the one or
more constraints
include a mirroring constraint for the first motion vector difference and the
second motion vector
difference, and wherein the first motion vector difference and the second
motion vector difference
have a same magnitude and a different sign.
[0363] Aspect 61. The method of any of Aspects 58 to 60, wherein the one or
more constraints
include a zero value constraint for at least one of the first motion vector
difference or the second
motion vector difference.
[0364] Aspect 62. The method of Aspect 61, wherein, based on the zero value
constraint, the
one or more refined motion vectors are determined using the selected motion
vector search strategy
by maintaining a first one of the first motion vector difference or the second
motion vector
difference as a fixed value and searching relative to a second one of the
first motion vector
difference or the second motion vector difference.
[0365] Aspect 63. The method of any of Aspects 55 to 62, wherein the selected
motion vector
search strategy is a bilateral matching (BM) motion vector search strategy.
[0366] Aspect 64. The method of any of Aspects 55 to 63, wherein the one or
more refined
motion vectors are determined based on one or more constraints associated with
the selected
motion vector search strategy, and wherein determining the one or more refined
motion vectors
based on the one or more constraints comprises: determining a first refined
motion vector by
searching a first reference picture around the first motion vector based on
the selected motion
vector search strategy; and determining a second refined motion vector by
searching a second
reference picture around the second motion vector based on the selected motion
vector search
strategy; wherein the one or more constraints include a motion vector
difference constraint.
[0367] Aspect 65. The method of Aspect 64, wherein determining the first
refined motion vector
and the second refined motion vector comprises: minimizing a difference
between a first reference
block associated with the first refined motion vector and a second reference
block associated with
the second refined motion vector.
96
CA 03216820 2023-10-13
WO 2023/278964 PCT/US2022/073155
[0368] Aspect 66. The method of any of Aspects 55 to 65, wherein the plurality
of motion vector
search strategies includes at least two of a multi-pass decoder side motion
vector refinement
strategy, a fractional sample refinement strategy, a bi-directional optical
flow strategy, or a sub-
block based bilateral matching motion vector refinement strategy.
[0369] Aspect 67. The method of any of Aspects 55 to 66, wherein the selected
motion vector
search strategy comprises a multi-pass decoder side motion vector refinement
strategy.
[0370] Aspect 68. The method of Aspect 67, wherein the multi-pass decoder side
motion vector
refinement strategy includes at least one of one or more block based bilateral
matching motion
vector refinement passes or one or more sub-block based motion vector
refinement passes.
[0371] Aspect 69. The method of Aspect 68, further comprising: performing the
one or more
block based bilateral matching motion vector refinement passes using a first
constraint associated
with at least one of a first motion vector difference or a second motion
vector difference; and
performing the one or more sub-block based motion vector refinement passes
using a second
constraint associated with at least one of the first motion vector difference
or the second motion
vector difference, wherein the first constraint is different than the second
constraint.
[0372] Aspect 70. The method of any of Aspects 68 or 69, wherein the one or
more sub-block
based motion vector refinement passes include at least one of a sub-block
based bilateral matching
motion vector refinement pass or a sub-block based bi-directional optical flow
motion vector
refinement pass.
[0373] Aspect 71: A method of processing video data in accordance with any of
the operations
of Aspects 37 to 70.
[0374] Aspect 72: A computer readable storage medium comprising instructions
that, when
executed by one or more processors of a device, cause the device to perform
the operations of any
of Aspects 37 to 70.
[0375] Aspect 73: An apparatus comprising one or more means for performing any
of the
operations of Aspects 37 to 70.
97
