Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
3 ~ 3~
BACKGRO~ND O~ T~E rNVEN~ON
This invention relates generally to television appa-
ratus, and more particularly to a method and apparatus for
digitally encoding and processing an NTSC color television
signal.
In digital television systems, it i5 necessary to
reduce the bit rate of the digital television signals, usuall~
pulse code modulated (PCM), whenever a transmission channel or a
digital store is limited in capacity~ One way of doing this i5
to lower the PCM encoding frequency, Fs; however, the Nyquist
sampling limit is soon reached, and further reduction ln sampling
frequency results in beat distortions due to "alias components"
when the lower sidebands of Fs overlap the baseband video fre~
quencies; since the baseband video bandwidth, Fv, for the NTSC
system of television is 4.2 MHz, the Nyquist sampling limit is
reached when Fs-2F , l.e., FX=8.4 MHz.
It is known from applicant's Pat~ No. 4,065,784 that
NTSC color television signals can be digitally encoded at sub-
Nyquist rates by placing the alias components into those parts of
~he spectrum not normally occupied by ~he luminance or chrominance
components of the video signal. In the patented system, Fs is
exactly 2FsC + 1/4Fh,or 2FSc + 1/4Fh, where FsC is the NTSC color
subcarrier frequency and Fh is the line-scan frequency. Most of the
~ .
aIias signals in the thus encoded signal are removed from the
baseband video by comb filtering between (FS-Fv) and Fv.
It has recently been proposed that the NTSC color
television signal be encoded at a sampling rate four times the
color subcarrier frequency (i.e., 4Fs~) in order to maintain
picture quality and ease digital processing. Encoding at this
sampling frequency results in a bi~ rate of 114 Mbs~ which may
be excessive for certain applications. Although ~ub-Nyquist
- 2 -
3~
encoding is a way t~o reduce the bit rate, the quarter line fre-
~uency of~set described in the aforementioned patent makes it
difficult and expensive to obtain the correct samples from a
4FSc encoded PCM television signal.
It is the primary object of the present invention to
provide a method and apparatus for obtaining a sub-Nyquist
sampled signal directly from 4FSc samples without the need for
interpola~ion, from which the 4F ~ samples can again be readily
reconstructed using comb filters to remove alias components.
Another object of the invention i~ to provide a method
and apparatus for obtaining from an analog NTSC color television
signal a sub-Nyquist encoded PCM signal ~rom which 4Fs~ samples
can be readily reconstructed.
SUMM~RY OF THE INVENq~ION : ::
: :
Briefly, according to one aspect of the invention, a
sub-Nyquist sampled NTSC color television signal having a
sampling frequency of twice the color subcarrier, that is, 2FSc,
in which the resultant alias components are interlaced between
the luminance and chrominance peaks o the energy spectrum to
allow their removal from the baseband video by comb filtering, is
obtained by retaining every other sample of a PCM-encoded signal -
sampled at four times the color subcarrier frequency (i.e., 4FSc)
and introducing one sample displacement every two sequential
television lines. This is accomplished without the need for
interpolation by introducing a 180 phase shift in the 2FsC
sampling frequency every alterna~e time sequential television line.
Using comh filterir.g to remove alias components, the 4FSC samples ~ -
can be reconstructed, and the process of converting 4FSc encoding
to sub-Nyquist and back to 4FSc can be repeated as necessary
without impairing the television picture beyond that introduced
during the first conversion ~ reconvexsion process.
- 3 -
3~
Acccrding t~ another aspect of the inventi~n, instead
of obtainin~ the sub-N~uist sampled signal from the 4FsC samples,
means are provided for sampling an analog NTSC color television
signal at a sampling frequency 2FsC to obtain a PCM-encoded
signal in`which the alias components are interlaced between the
luminance and chrominance peaks of the energy spectrum. By
using comb filtering to remove alias components, 4FSc samples
can be constructed from the 2FSc samples, and the thus-produced
4FSc samples can be sampled in the manner described in the pre~
ceding paragraph, if desired, to convert back to 2FSc samples.
More particularly, there is provided:
A method of deriving a sub-Nyquist sampled pulse code
modulated NTSC color television signal from a pulse code modu-
lated color television signal sampled at 4FSc, where FSc is the ~;~
color subcarrier frequency, comprising the steps of:
selecting every other 4FSc sample in each line of the 4FSc
sampled television signal; and
intro~ucing a one 4FSc sample displacement every two sequen-
tial television lines.
There is also provided:
Apparatus for deriving a sub-Nyquist sampled pulse
code modulated NTSC color television signal from a pulse code
modulated color television signal sampled at a frequency of 4FSc,
where FSc is the color subcarrier frequency of the color tele-
vision signal; said Rpparatus comprising~
means for selecting every other 4FSc sample in each line of
the 4F5c sampled television signal; and
means for introducing a one 4FSc sample displacement every
two sequential television lines.
There is further provided:
_ 4 -
,s' . '':
~ ~3~
Apparatus for deriving a sub~Ny~uist sampled pulse ;~
code modulated NTSC color television signal ~rom a pulse code
modulated NTSC color television signal sampled at a rate of 4F5c,
where FSc is the fre~uency of the subcarrier wave of the tele- :
vision signal which is modulated with chrominance information~
and where the samples are taken at points about 90 of said sub~
carrier wave away from each other in each scanning line, the :
frequency of which is Fh, said apparatus comprising~
an AND gate having first and second inputs connected to :
receive at its first input the 4FSc sampled television signal,
and
means connected to receiVe the 4FsC sampled television signal
and producing and applying to the second input of said AND gate : :
control pulses to cause said AND gate to select every other :~
sample in each line of the 4FSc sampled television signal, where- ~ :
by the selected samples are at points about 180 of said sub- ~ ~:
carrier wave away from each other, and also to displace the ~
selected samples ~y about 90 of said subcarrier wave every two ~ :
sequential television lines.
BRIEF DESCRIPTION OF THE DRAWINGS
Other objects, features and advantages of the inven-
tion will become apparent, and its construction and operation
better understood, from the following description, taken in con-
junction with the accompanying drawings, in which:
FIG. 1 is a diagram showing the foldo~er of the lower
encoded video sideband onto the baseband video when Fs is less
than 2F~; ~ .
FIG. 2 is ~ diagram showing the spectral characteris-
tic of the main Y and C energy components of an NTSC color tele-
vis~on signal within the chrominance sidebands;
- 5 -
36Z6~
FIG. 3 is a diagram showing desixed spectral charac- ~
.
teristics for a sub-Ny~uist sampled NTSC color television signal;
FIGS. 4, 5 and:6 are block diagrams of three different
forms of comb filters useful in the practice of the invention;
FIG. 7 is a diagram showing the frequency response of ~-
one form of comb filter;
FIG. 8 is a diagram showing the -frequency response
of another form of comb filter; ~ ~
FIG. 9 is the spatial sampling pattern on the image ~ .
plane of a sub-Nyquist sampled PCM NTSC color television signal
derived from a 4FSc sampled signal;
FIG. 10 is a block diagram of a system for sampling
an analog NTSC television signal at a sub-Nyquist frequency 2FSc;
FIG. 11 is a set of waveforms at different points in
the system of FIG..10 useful to understanding its operation;
FIG. 12 is a block diagram of a system or deriving
a sub-Nyquist sampled :signal from a 4FSc sampled signal;
FIG. 13 is a block diagram of the equivalent digital
implementation of the filter shown in FIG. 4; and
FIG. 14 is a block diagram of a digital-to-digital
converter for converting from sub-Nyquist to 4FSc. ~ :
DESCRIPTION OF' THE PREFERRED EMBODI~!ENT --
Before proceeding to the description of the sub-
Nyquist encoding system according to the invention, it will be
useful to review the problem inherent in sub-Nyquist encoding of
television signals, and the significant spectral characteristics
of the NTSC color television signal. As was mentioned earlier~
whsn in PCM encoding NTSC television signals it is attempted to ~ ~
reduce the sampling frequency, Fs, below the Nyquist limit, ~ ~ :
beating or "aliasing" distortion occurs due to the lower side-
bands of Fs overlapping the baseband video signals, as shown in
'3~
3~
FIG. 1. For NT~C teleyisionl Fv~ 4.2MHz; conse~uently the Nyquist
sampling limit is reached when Fs~ 2FV, or when Fs= 8.~MHZ.
In the NTSC color tele~ision sign~l, the spectral
energy of the luminance (Y) signal is essentially centered at
harmonics of the line scanning fre~uency Fh; i.e., nFh, where n
is an integer~ The chrominance (C) signal spectral energy peaks
at odd harmonics of l/2Fh; i.e., (n~ 1/2)Fh. Thus, the luminance
and the chrominance energy bundles are frequency interleaved as
shown in FIG. 2. `
To encode the PCM NTSC color television signal at a
sub-Nyquist rate and subsequently remove the alias components,
the encoding frequency Fs should be chosen to frequency interlace
the alias components between the desired luminance and chromi~ -
nance components, as shown in FIG. 3. Included are the peak
baseband frequency.components o the luminance, Y~, and chromi~
nance, CB, signals, and the alias luminance components, YA~ and
alias chrominance components, CA The sub-Nyquist system
described in the aforementioned patent uses an encoding frequency
of 2FSc I 1/4Fh to produce the frequency spectrum shown in FIG. 3.
Unfortunately, such samples are not readily obtainable from a
4E'Sc encoded PCM television signal that may become the recom-
mended standard in the broadcasting industry.
Applican~ has shown in his aforementioned patent that
when the sub-Nyquist sampling fre~uency selected for NTSC color
television signals produces the spectral frequency response
shown in FIG. 3, the luminance and chrominance information can
be recovered and the undesired alias components rejected by means
of suitable transversal comb filtering. However, for clarity of
understanding of the present method for PCM-encoding NTSC color -
television signals at a sub-Nyquist rate, the technique for
rejecting alias signals by means of comb filtering will be
- 7 -
~, ~
3~
briefly reviewed. As seen in ~IG. 3, the centers of the alias
energy bursts are sepaxated by frequency intervals of l/2Fh.
The comb filter, therefore, should have its maximum responses
(teeth) or minimum responses (nulls) at frequency intervals of
1/2Fh. Such comb filter can be made by combining videQ signals
from alternate time-se~uential television lines. For example,
in a particular field, line ~ would be combined with line (~ -2)
or ( R +2~. Since a transversal filter that combines television
lines can cause objectionable transients and a loss of vertical
resolution in the television picture, it is desirable to combine
as few lines as possible to obtain the necessary comb filter
frequency response. The alias components can be removed from
a sub-Nyquist encoded NTSC color television signal having the
spectral frequency response shown in FIG. 3 by using either of
the following com~ filter algorithms:
1. Add television line Q to ~-2).
2. Add television line ~ to ( Q +2); or
3. Add television line ~ to 1/2~ 2)~(~ +2)~.
All lines are from a single field to avoid the need for inter-
field storage in the decoder. Block diagrams of three possiblecomb filters are shown in FIGS~ 4, 5 and 6, which respectively
correspond to FIGSo 6, 7 and 8 of the aforementioned patent.
The frequency response of the filters of FIGS. 4 and 5 is shown in
FIG. 7, and the frequency response of the FIG. 6 filter is shown
in FIG. 8. It will be noted in both FIGS~ 7 and 8 that there are ~ -
~nulls at the alias peak amplitude frequency components, and the
teeth are centered at the peak amplitude frequency components of
the baseband signals.
As has been previously noted, the sub-Nyquist sampling
frequency selected for NTSC color television signals should result
in the spectral frequency response shown in FIG. 3 in order
,,
` ~13~Z~
:.
readily to recover the luminance and chrominance information
and reject the undesired alias components, A sub-Nyquist sampling
frequency of nFh will result in luminance alias components over-
lapping the baseband main luminance components and chrominance
alias components overlapping the baseband chrominance. Sub-
Nyquist sampling at nFh with 1/2Fh offset will cause luminance
alias components to overlap the baseband chrominance, and the
chrominance alias components to overlap the baseband luminance.
Thus, these two sampling methods will not work. In the system
of the aforemen~ioned patent, a l/4Fh offset in the nFh sub-Nyquist
sampling rate places the alias components between the baseband
luminance and chrominance main frequency components, which
operates satisfactorily. However, the quarter line frequency
offset makes it dlfficult and expensive to obtain the correct
samples from a 4FSc encoded PCM television signal, which may
become the recommended standard in the broadcasting industry, thus
presenting the need for readily obtaining the sub-Nyquist sampled
signal from a 4FSc sampled signal. The fact that 4FSc-9lOFh
suggests that the sub-Nyquist sampling frequency should be 2FSc,
or 455Fh, which dictates that means other than a l/4Fh frequency
offset must be provided to interlace the resultant alias com-
ponents between the luminance and chrominance energy peaks.
This is accomplished according to the present inven-
tion by introducing a 180 phase shift in the sampling frequency
every alternate time sequential television line. The effect of
this will be seen from examination of FIG. 9 which depicts the
pattern of the spatial picture samples on the image plane, the
dots represen*ing sampling points taken at the rate of 14.32MMz,
or four times the color sub-carrier frequency, which are
spatially aligned on successive video scan lines ~Q~
~, (Q ~1~, etc., and spaced 41F , or 70 nanoseconds apart. The
_ g _
. ~
circles represent sa,mpling points of the sub~Ny~uist sampled PCM ;'
NTSC color teIevision signal derived from the'4Fsc sampled signal;
it is seen that the circles are spread apart 2F' ~ or nominally
140 nanoseconds, along each scan line, and are displaced one ~'~
sample interval ever~ two sequential television lines. That is,
the circles in lines (Q -1~ and Q are displaced to the right one
sample interval with respect to the circles in the two television ~ -
lines preceding and following them. In summary, the sampling
frequency of the sub~Nyquist samples is exactly 2FSc with appro-
priate phase shifts at the starts of different television lines.
It will be appreciated tha~ with this technique the sub-Nyquist ~ ,
encoded signal can be obtained from the 4F$c PCM encoded signal
without any need for interpolators; the process is one of simply
retaining the appropriate ones of the 4F5c samples. With proper
precautions it is possible to go back and forth between the 4FSc
encoded signal and the 2FSc encoded signal an unlimited numher
of times without any further video degradation than that caused ~i'
by the first translation from 4FSc to 2FsC, as long as the same
samples from the same frames are selected. This guarantees that
the sub-Nyquist digital stream will always consist of a selected
set (as per FIG. 9) of unprocessed original 4FsC samples.
It i5 interesting to note that if it is assumed that
no two samples in the sub-Nyquist digital stream are separated
by less than the period l/2Fsc, the resultant long term average
sampling frequency is 2Fsc-l/4Fh~ Thus, there is a 1/4Fh fre-
quency offset, except that here, unlike in the patented system~
this frequency offset is not continuous but is a result of the
sample displacements that are introduced every two television ;~
lines.
FIG. 10 shows in block diagram form a system for PCM-
encoding an NTSC color television signal at the sub-Nyquist rate
-- 10 --
~., :
~.~36~f~ :
' '` ' '
o~ 2FSc. An input NTSC video signal in analeg form, which has
preferably been subjected to comb filtering for the reasons and
in the manner discussed hereinafter, is received on line 30 and
coupled to the input of an analog-to-digital converter 32~ to ~:
sync stripper 34, and to a color subcarrier regenerator 36. The -
color subcarrier regenerator is of conventional design and
regenerates the color subcarrier which, in the NTSC color tele- -
vision system, is 3.5BMHz. The regenerated color subcarrier
signal is applied to and con~rols a clock generator 38 which
produces two trains of clock pulses both having a frequency of
4FSc, but with one 180 out of phase with respect to the other,
as shown by waveforms (A) and (B) of FIG. ll. The 4F5c pulse
train is applied as one input to an AND circuit 40 and the 4F
pulse train is applied to the clock input of a "D-type" flip-flop
circuit 42, the Q terminal of which is connected as a second
input to AND circuit 40. The sync stripper 34, of conventional
design, p.roduces horizontal drive pulses (waveform C of FIG. 12) ~.
in synchronism with the horizontal sync pulse of successive
television lines, which are applied to the clock input of each of
two flip-10p circuits 44 and 46, both of which are of "D-type".
The Q output of flip-flop 44 is connected to the data (D) input ~:
so as to produce a pulse signal at half the horizontal drive . `
frequency, illustrated in waveform (D) of FIG. 12, which is :~
applied to the clock terminal of a fourth flip-flop circuit 48.
The Q output of flip-flop 48 is connected to its data input, and ~
'::
the Q output (waveform E) is applied as one input to an AND
circuit 50, the other input to which is the Q outpu~ of flip-flop
42 (waveform G). The Q output of flip-flop 42 (waveform F) ~:
resets flip-flop 48. The output of AND circuit 40, the result of
ANDing waveforms (A) and ~G), shown as waveform (H) of FIG. 11,
is applied to the clock terminal o:E a fifth D-type flip-flop 52
- 11- , .`~
-
~3~
the Q QUtpUt terminal of which is connected to it~ data terminal. -
With the described connections, flip-flop 52 produces at its Q
output the clock signal shown as waveform (I) of FIG. 12 having
a frequency of 2F5C, which is applied to ADC 32. The ADC, which
in the present embodiment is a PCM encoder, samples the analog
video signal under control of this clock signal, which shifts
sampling phase every 2H, or every alternate television line;
this gives the sampling pattern shown by the circles in FIG. 9O
The balance of the circuit illustrated in FIG. 10 is
provided to minimize flicker at vertical color transitions by
~orcing the first sampling points of correspondin~ lines on odd
fields to be superimposed, and the sampling points of all corres- ~
ponding lines on even fields to be superimposed. This relation- -
ship is guaranteed in the system of FIG. 10 by resetting of the
clock phase at the start of each frame. This is accomplished by
applying successive frame pulses, from the sync stripper 34
to the clock terminal of a flip~flop 54, the data input terminal
of which is set "high", and the "reset" input of which is con-
nected to the Q output of previously mentioned flip-flop 46. The
Q output of flip-flop 54 is applied as one input to an AND circuit
56, the other input to which is the output of an AND circuit 58
having the H-dri~e signal (waveform C) and the Q output of flip~
flop 44 as inputs. The output of AND circuit 56 is applied to
the "set" terminal of flip-flop 46, the data input of which is
set "low", and to the clock terminal of which is applied the H-
drive signal. With the described connections, the Q output of
flip-flop 46 applied to the "set;' terminal of flip-flop 44
ensures that ~he same spatial samples are selected on adjacent
frames.
It is important to the proper operation of the
encoding system of ~IG. 10 that the analog NTSC color television
- 12 -
~; :
. . . .,, ,... " . .. .. . . .
~3~2~
' '
signal be free of enexgy components at frequencies ~n~l/4)F
within the spectrum extending from (2~Sc~Fv) to Fv/ because
otherwise the alias components generated by sub-Nyquist encoding
would overlap the baseband video siynal main spec$ral energy -
components and would be inseparable. This problem can be avoided
by comb filtering the teIevision signal prior to sub-Nyquist
encoding tv remove all (n~l/4)Fh energy components, with care ~`
exercised, however, to use the proper type of comb filter in
order to avoid excessive loss of vertical resolution. If pre-
encoding comb filteriny is employed, one should use only the two~
line comb filters shown in FIGS. 4 and 5, using either prior to
sub-Nyquist encoding and the other for post-encoding filtering.
FIG. 12 shows in block diagram form a system for
obtaining a signal at the sub-Nyquist rate of 2FSc from an NTSC
color television signal PCM encoded at 4FSc rate. A digital NTSC
color video signal, PCM encoded at 4FSc, which has preferably ;~,
been subjected to comb filtering for the reasons and in the manner ~ `
previously discussed, i9 received on line 31 and coupled to the
:. . .
input of ~ND gate 33, sync stripper 35l and 4FSC clock regenerator
39. It should be noted that the PCM signal at line 31 could
consist of either a serial bit stream at 4Fscx N bit rate, where
N is the numbar of bits used to quantize sach video sample, or
a 4FSc, N parallel bi~s, PCM encoded signal. In the latter case,
line 31 would actually consist of N lines and gate 33 would
represent N AND gates. Clock regenerator 39 produces two trains -~
of clock pulses both having a frequency of 4FSc but with one 180
.
out of phase wi~h respect to the o~her, as shown by waveforms
(A) and (B~ of FIG. 11. ~;
The digital sync stripper 35 performs analogous
unctions to those of sync stripper 34 of FIG. 10. The AND gate
33, controlled by gating signal (I) of FIG. 11, allows every
., .
~ - 13 -
,~ ~ '.'`.
. . . . - - . - . . . .. ... i . . .
3~2~L
other 4FSc sample to pass through; in addition, it pxo~ides the
extra sample displacement every 2H, ox every alternate TV line,
to give the sampling pattern shown by the circles of FIG. 9.
All of the other circuits of the system of FIG. 12
are the same circuits as in FIG. 10, and the waveforms of ~IG. 11
and their description given in the ~peration of the circuits of
FIG. 10 also apply to FI~o 12
Whether the sub-N~quist sampled signal is derived
from 4FSc samples using the system of FIG. 12, or obtained by
10 sampling of an analog NTSC video signal using the system of FIG~ ~;
10, 4FSC samples may be regenerated in a number of ways, one ~`~
example of which is shown in FIG. 13. This is a digital imple- .
mentation of the filter circuit of FIG. 4 for sub-Nyquist to
super-Nyquist digital-to-digital conveXsion. The sub-Nyquist
signal~ at a frequency of 7.16Mw/sec. is xeceived on line 60 and
applied to a digital delay device 52 having a delay of two tele-
vision lines, to ~ne terminal 64a o~ a switching device schema-
tically shown at 64/ and also to the input terminal of a low-pass
filter and interpolator 66. The switching device 64 has a
switching rate of 4FSc thereby to alternately switch between
undelayed video samples (which may be designated line~ ) and
video samples delayed by two television lines (that is, from
line ~-2). This operation of adding the digital bit streams from
two television lines (the comb filter) really consists of insert
ing the current samples of line ~ between the video samples from
line ( Q -2) whereby the 4FSc sampling rate is reobtained. The
resulting signal, however, will exhibit a comb filter character-
istic throughout the whole video baseband; this problem is over-
come by applying the 4FsC signal to a high pass filter 68 having
characteristics to limit the comb filter response to the fre-
quency band above (2FSC-Fv). The non-comb-filtered lower video
''` :
... . . . . . .
L3~,Z~
baseband is obtained b~ lo~-p~ss ~iltering the signal of line Q ~ :
from dc to (2FSc-Fv) in low pass filter 66 and doubling its
sampling rate by means of a linear phase interpolator of known
construction. The output of filter and interpolator 66 is added
in a summing device, diagrammatically shown at 70, to the high
passed signal from filter 68 to construct the super-Nyquist
sampled video signal which, in this case, has a sampIin~ frequency ; ~ ;
of 4FSc or a sampling rate of 14.3 Mw/sec.
It has been shown in BBC Research Department Report ~;
10 1977/21 entitled "Digital Video: Multiple sub-Nyquist coding" ~ ; -
by J. H. Stott and T. J. Phillips, that repeated sub-Nyquist
encoding of PAL television signals does not significantly impair ;~
the television picture beyond that introduced by the first
sampling operation. Since the herein described sub-Nyquist
technique for an NTSC color television signal can satisfy all the
requirements set forth in the 3BC report for a PAL signal under-
going multiple sub-Nyquist encoding, the analysis given therein
for the PAL signal can be readily extended to the NTSC signal to
prove that no extra degradation due to re-sampling occurs.
Indeed, it can be readily shown that the re-sampling process can
be performed indefinitely without any degradation beyond that
introduced by the first conversion. This will be more readily
understood from consideration of the block diagram of FIG. 14,
which is of the block diagram of FIG. 13 re-drawn to its exact
equivalent. In the arrangement shown in FIG. 14, the television
line (~ -2) delayed by 2H by delay device 74, is high pass
filtered above (2FsC-FV) and i5 added in summing device 80 to
the non-delayed television line Q which is low-pass filtered to
(2F~C-Fv). The low pass filter includes an interpolator for
generating video samples in television line ~ that are time
coincident with the samples from television line (Q -2). The
' ~`;`,'
- 15 -
';
: ~3~2~ ~
nondelayed digital bit stxeam ~of line ~ is also fed directly to
an output switch 82 having a switching rate of 4Fsc or 14.3MHz.
The super-Nyquist signal is regenerated at switch 82 by taking
the essentially unprocessed samples from line ~ (which are also
original samples of the 4FSc sampled video~ and inserting between
those samples the low-passed/high-passed combination of ~ and
( R -2) samples. The resultant super-Nyquist signal is truly a
4FSc PCM NTSC color television signal which has undergone a comb
filter process equal to that shown in FIG. 4. It will now be
clear that one can repeat the sub-Nyquist coding process indefi-
nitely as long as the same samples from the same frames are
always selected, since this guarantees that the sub-Nyquist
digital stream will always consist of a selected set of unprocessed
4FSc samples.
Tests of this sub Nyquist sampling system with a
variety of NTSC color television signals have good results. Using
the system of FIGo 10 or FIG~ lZ to obtain a sub-Nyquist sampled
signal, the 4FSc samples were regenerated using the system of ~,
FIG~ 14. The results were better than those obtained using the
20 2FSc ~ 1/4Fh or 2FSc ~ 1/4Fh sampling rates taught by applicant's ~;
Pat. No. 4,06S,784. The improvement results from the elimination
of the 7.5Hz flicker at vertical color transitions by the use of
the frame resetting pulse that forces the sampling points on
adjacent frames to be superimposed. If the frame pulse from the
sync stripper is not used, results equivalent to those obtained ;~
with the previous system are obtained. ~ ;
It should be noted that the effect of sub-Nyquist
sampling on the luminance details is dependent on the angle
between the picture detail and the scanning lines. Vertical
luminance transitions are normally not impaired since their fre-
quency components do not extend into the passband of the comb
' :i
- 16 -
i .... . ...
~36;~i
filtex. With dia~onal t~ansitions, however, the comb filter
reduces the amplitude of wanted-frequency components above fs-fv,
and the corresponding alias components are not completely re- -
moved. In general~ however~ a small deterioration of diagonal
transitions does not appear subjecti~ely objectionable. Hori~
zontal transitions are vir~ually unaffected because their energy
components do not normally fall within the nulls of the comb
filter.
., :.
The effect of the sub-Nyquist system on chrominance
was judged imperceptible with most broadcast signals. Only highly
saturated colors generate one ox two lines of wrong chrominance
at sharp vertical color transisitions. This problem is parti-
cularly noticeable with 100% saturated split field color bars.
Other stationary pictures including scenes from slides No. 1 to -`-
15 of the SMPTE Television Color Reference slide set were judged
not objectionably impaired.
' ' '`
'
; '
`
- ~7 - ~
. , , ` , , . '
