Note: Descriptions are shown in the official language in which they were submitted.
'~ Z~
CA 73,993
SIGNAL PROCESSING CIRCUIT
HAVING_ A NON-LINEl~R TRANSFER FUNCTION
S This invention rela-tes to an electrical circuit
having a non-linear ampli-tude transfer characteristic, and
more par-ticularly, -to such a circui-t for selec-tively
providing amplitude restoration, peaking and at-tenuation
of picture vertical detail information in a color television
receiver including a comb filter or -the like for separating
the luminance and chrominance components of a color -tele-
vision signal.
In a color television system such as -the sys-tem
developed by the United States, the luminance and chrominance
components of a color -television signal are disposed within
the video frequency spectrum in frequency interleaved
relation, with the luminance componen-ts at integral
multiples of the horizontal line scanning frequency and the
chrominance component at odd multiples of one-half the line
scannin~ frequency. Various comb filter arrangements for
separating the frequency interleaved luminance and chromi-
nance components of the video signal are known, for
example, from U.S. Patent 4,143,397 (D. D. Holmes) and
U.S. Patent 4,096,516 (D. H. Pritchard) and the references
cited therein.
A combed luminance signal which appears at the
luminance output of the comb filter has been subjected to a
"combing" effect over its entire band. The combing action
over the high frequency band portion which is shared with
chrominance signal components has the desired effect of
deleting chrominance signal components~ Extension of this
combing action into the low frequency band portion which
is not shared with the chrominance signal components,
however, is not needed to effect the desired removal of
36 chrominance signal components, and serves only to
unnecessarily delete luminance signal components.
Components in the lower end of the unshared band which are
subject to such deletion are representative of "vertical
detail" luminance information. Preservation of such
''~k
.
,~
., .
Z8
1 - 2 - ~CA 73,993
vertical detail :is desirable to avoid loss oE vertical
resolution in -the luminance con-tent of a d:Lsplayed image.
One arrangement for preserviny the vertical
detail information employs a low pass fil-ter coupled to -the
output of the comb filter at which the "combed" chrominance
component appears. The upper cut-off frequency o~ this
filter lies below the band occupied by the chrominance
signal componen-t (with an illustrative choice being just
below 2 MHz). The filter selectively couples signals below
the chrominance band from the chrominance output of the
comb filter to a combining network where the selec-tively
coupled signals are summed with combed luminance output
signals from the comb filter. The combined signal includes
a "combed" high frequency portion (occupying a band of
frequencies above the filter cut-off frequency) from which
chrominance signal components have been removed, and an
uncombed (i.e., "flat") low frequency portion in which all
luminance signal components have been preserved.
It is sometimes desirable to enhance or peak the
vertical detail inEormation of a displayed image by adding
back to the luminance signal a greater amount of the
vertical detail signal than is required to restore the
luminance signal to its original form (i.e., a "flat"
amplitude characteristic). The additional vertical detail
signal then serves to emphasize vertical detail information
so as to enhance picture detail resolution. For low level
luminance signals, however, such enhancement tends to
30 produce objectionable visible effects when noise inter-
ference is present and undesirably enhanced along with the
vertical detail information of the luminance signal.
Also in this instance, alternate line set-up
variations (ALSUV) when present in the video signal are
35 also undesirably enhanced. The ALS W phenomenon is a
form of low level signal interference manifested by
variations in the black level of the video signal from
line-to-line, and may be caused by misalignment of signal
processing systems at the broadcast transmitter, for
128
1 - 3 - RCA 73,993
example. The ALSUV interference is par-ticularly noticeable
for low level video signals of about five percent of the
maximum expected video signal amplitude, and produces
objectionable visible effects on a reproduced image which
are undesirably magnified when vertical detail enhancement
is provided.
A technique for minimizing the adverse effects of
noise and other undesirable components of a video signal
employs a process commonly referred to as signal "coring,"
wherein small amplitude excursions of the signal
(including noise) are removed as described in U.S.
Patent 3,715,477, fvr example.
One advantageous system which accomplishes coring
of the vertical detail signal in a manner which does not
impair (e.g., "smear") vertical detail information,
particularly with regard to low level detail signal
information which is to be restored to the luminance signal,0 is described in United States Patent No. 4,223,339
of W. A. Lagoni and J. s. Fuhrer entitled
"Video Image Vertical Detail Restoration And Enhancement ".
The system described
therein also advantageously provides for enhancement of
the vertical detail information sub`stantially without
simultaneously enhancing interferin~ signal components
such as noise and alternate line set-up variations.
A system wherein large amplitude vertical detail
signals are pared (amplitude reduced or attenuated~ to
30 prevent kinescope "blooming" which would otherwise distort
or obscure detail information is disclosed in United States
Patent Mo. 4,245,238 of J. S. Fuhrer
entitled, "Non-Linear Processing Of Video Image Vertical
Detail Information ".
3S Consistent with the techniques described in the
last-mentioned copending patent applications, in accordance
with the principles of the present invention there are
disclosed herein signal processing circuits having a non-
linear amplitude transfer function for selectively providing
,, ~ . - .
828
CA 73,993
restoration, enhancemen-t (peaking), and paring for small,
moderate, and large amplitude video signals, respectively.
A circuit in accordance with the present
inven-tion comprises an amplifier having an input terminal
and an output terminal, including an amplifier device with
an input and an outpu-t. A first feedback path including a
first impedance is coupled between the input and output
of the amplifier device. ~ second feedback path including
a second impedance and a threshold switching network is
also coupled between the output and input of the amplifier
device. The switching network has an input coupled to the
output of the amplifier device, and an output coupled to
the output terminal of the circuit. The switching network
exhibits one conductive state in response to signal
amplitudes of first magnitudes within a firs-t range, and
another conductive state in response to signal amplitudes
of second magnitudes greater than the first magnitudes
~0 within a second range.
In accordance with a feature of circuits according
to the invention, a third feedback path is provided between
the output and input of the amplifier device. The third
path includes a third impedance and an additional threshold
switching network with an input coupled to the output of
the switching network in the second feedback path. The
additional switching network exhibits one conductive state
in response to signal amplitudes of the first and second
magnitudes, and another conductive state in response to
signal amplitudes of third magnitudes greater than the first
and second magnitudes within a third range.
According to a further feature of circuits accor-
ding to the invention, the second feedback path also includes
an additional impedance. The switching network in the second
35 path is coupled to the additional impedance for switchably
controlling the current conduction through the additional
impedance.
In accordance with another feature of the
invention, circuits according to the principles of the
, . .: . . ,~ .
-
38~3
1 - 5 - RCA 73,993
present invention are employed in a color -television
receiver or similar system for -translatiny vertical image
detail information signals wi-th a non-linear transfer
function with respect to prescribed ranges of vertical
detail signal amplitude.
In the drawing:
FIGURE 1 illustra-tes a block diagram of a portion
of a color television receiver employing a non-linear
signal processor according to the present invention;
FIGURE 2 shows one embodiment of a non-linear
signal processing circuit according to the present
invention;
FIGURE 3 depicts a modiEied version of the
circuit shown in FIGUR~ 2;
FIGUR~S 4-6 depict amplitude response charae-
teristics which are useful in understanding the operation
of eircuits according to the invention; and
FIGURE 7 shows another embodiment of a circuit
according to the invention.
In FIGURE 1, a source of composite color video
signals 10 ineluding luminanee and ehrominance components
supplies video signals to an input of a comb filter 15
of known configuration, such as a comb filter employing
charge transfer devices (CCD's) as shown in U.S. Patent
4,096,516. The luminance and chrominance components are
arranged within the video signal frequency spectrum in
frequency interleaved relation. The luminance component
has a relatively wide bandwidth (extending from D.C. or
zero frequency to about four megahertz). The upper
frequency range of the luminance component is shared with
the chrominanee component, which comprises a subcarrier
signal of 3.58 ~5Hz amplitude and phase modulated with
color information. The amplitude versus frequency
response of comb filter 15 with respect to luminance
combing action exhibits a peak amplitude response at
integral multiples of the horizontal line scanning
frequency (approximately 15,734 Hz), extending from D.C.
or zero frequency, and an amplitude null at odd multiples
3~8
1 - 6 - RCA 73,993
of one-half the line scanning frequency, including -the
3.58 MHz chrominance subcarrier frequency. The amplitude
versus frequency response of comb filter 15 with respect
to chrominance combing action exhibits a peak amplitude
response at odd multiples of one-half -the line frequency
including 3.58 MHz, and an ampli-tude null at in-tegral
multiples of the line frequency.
A "combed" luminance signal (Y) from the luminance
output of comb filter 15 is coupled via a low pass filter
22 to one input of a signal combining network 30. Fil-ter 22
is arranged to pass all luminance signals below a cut-off
frequency of approximately 4 MHz, and serves to remove
noise and clock frequency components of switching signals
associated with the switching operation of comb filter 15
when o~ a CCD type comb filter.
A "combed" chrominance signal (C) from the
chrominance output of comb filter 15 is applied to a
chrominance signal processing unit 64 for generating R-Y,
B-Y and G-Y color difference signals, and to an input of
a low pass vertical detail filter 35. Unit 64 includes a
suitable filter for passing only those signal frequencies
from comb filter 15 which occupy the band of chrominance
signal frequencies. ~ilter 35 exhibits a cut-off frequency
of approximately 1.8 MHz, and selectively passes those
signal frequencies present in the combed chrominance
signal output of comb filter 15 which lie below this
cut-off frequency. Signal frequencies in this region
represent vertical detail luminance information which is
absent from the combed luminance signal and which must be
restored to the luminance signal to avoid loss of vertical
resolution in the luminance content of a displayed image.
Such vertical detail restoration as well as vertical detail
enhancement and paring is accomplished as follows.
Vertical detail signals from the output of
filter 35 are supplied to a non-linear signal processing
circuit 50 which will be described in detail subsequently.
The amplitude transfer characteristic of signal processor 50
3~8
1 - 7 - RCA 73,993
is illustrated by FIGURE 5. The fo]lowing remarks
concerning the response to positive (+) polarity signals
also apply -to signals of negative (-) polari-ty.
The signal processing circuits wi-thin processor 50
produce a signal amplitude transfer (gain) characteristic,
as shown in FIGURE 5, for three regions I, II and III with
respect -to three predetermined ranges of vertlcal detail
signal amplitude. A prescribed gain response is produced
in a restoration region I for low level signals ~e.g.,
signal amplitudes of about five percent of maximum expected
amplitude), so that low level detail signals along with
noise and other undesired componen-ts are processed without
16 enhancement in region I. The peak amplitude of vertical
detail signals of moderate amplitude (e.g., signal
amplitudes between about five percent and forty percent of
maximum expected amplitude) are processed within enhancement
region II with a gain of approximately three, for example,
to thereby emphasize the vertical detail information and
enhance picture definition in this region. The peak
amplitude of relatively large amplitude vertical detail
signals (e.g., between about forty percent of maximum
expected amplitude and maximum amplitude) corresponding
to high contrast im~ages such as lettering, for example,
are reduced in amplitude or "pared" as indicated by the
amplitude response in region III, to avoid excessive
contrast and to prevent kinescope "blooming" which would
otherwise distort or obscure picture detail.
It is noted that in region I (vertical detail
restoration), low level vertical detail signal inEormation
has been restored in an amount sufficient to preserve
normal low level vertical resolution in the luminance
content of a displayed image. In this example, and as will
35 be seen hereafter, a prescribed restoration gain of
approximately two is imparted to small signal amplitudes
processed in region I. The gain in region I preferably is
that amount of signal gain which, in a given system, is
required to restore small amplitude excursions of the
.
CA 73,9~3
vertical det~il signal to the ]uminance signal so -that an
ultimately reconstituted luminance signal exhibits an
essen-tially "flat" ampli-tude response wi-th respec-t to
small amplitude detail signals. In -this connection it is
noted that the magnitude of the res-tora-tion gain is a
function of various factors, including the signal trans-
lating characteristics of networks coupled between the
outputs of comb filter 15 and a luminance processor 32
which processes ultimately reconstituted luminance signals,
and the relative magnitudes of the signals appearing at
the outputs of comb filter 15, for example.
The choice of the restoration gain as provided
by the amplitude transfer response for region I also
involves considerations of what results are acceptable
in a given video signal processing system. For example,
if the restoration gain is excessive, low level ALSUV
signal interference is likely to be visible. If the
~0 restoration gain is insufficient, significant combing
effects (i.e., signal peaks and nulls at different
frequencies) will appear in the vertical detail frequency
region below 2 MHz, resulting in lost low level vertical
detail information. Thus the slope of the amplitude
transfer characteristic in region I corresponds to the
amount of signal gain necessary to produce a desired
response (e.g., a flat luminance response) without
introducing unacceptable side effects. The signal
amplitude response for region I preferably exhibits a
fixed relationship with the response of the signal path
which couples the combed luminance signal (Y) from the
output of the comb filter 15 to combiner 30.
In region II (vertical detail enhancement), an
appropriate amount of vertical detail enhancement has been
35 provided by imparting additional gain to signals of
moderate amplitude in a manner which is considered to
benefit vertical resolution of a displayed image. As will
be seen hereafter, although the peak amplitude excursions
of moderate amplitude signals subject to enhancement are
1 - 9 - RCA 73,993
amplified with a gain greater than the restoration gain
in this example, small amplitude excursions thereof are
processed with the restoration gain (i.e., without
enhancement). Also, small amplitude signals not subject
to enhancement are processed with the restoration gain.
Thus enhancement of undesirable low level signal components
including noise and ALSUV interference is essen-tially
eliminated or reduced to an acceptable minimum, and image
"smear" of low level vertical detail information is
avoided.
The processed vertical detail signal from
processor 50 is summed in network 30 with the combed
luminance signal (Y) supplied via filter 22. The output
signal from combiner 30 corresponds to a reconstituted
luminance component of the video signal with the vertical
detail information thereof restored (region I), enhanced
(region II) and pared (region III) as discussed. The
reconstituted luminance component is afterwards coupled to
a luminance signal processing unit 32. An amplified
luminance signal Y from unit 32 and the color difference
signals from chrominance unit 64 are combined in a matrix
68 for providing R, B, and G color image representative
output signals. These signals are then suitably coupled
to image intensity control electrodes of a color
kinescope 70.
Referring now to FIGURE 2, there is shown one
circuit embodiment of non-linear signal processor 50.
Output signals from detail filter 35 are supplied as an
input signal (Si) via a coupling capacitor 72 and an
input resistor 73 to an inverting input of an operational
amplifier 75 included in processor 50. A non-inverting
input of amplifier 75 is coupled to a point of reference
35 potential (e.g., ground).
A first feedback network coupled between the
output and inverting input of amplifier 75 comprises a
feedback resistor 76. A second feedback path includes
the parallel combination of coring diodes 81 and 82 with
8~8
1 - 10 ~ RCA 73,993
a resistor 87, a feedback resistor 78, and a coupling
capacitor 77. Diodes 81 and 82 are mutually arranged to
conduct in response to opposite polarities of a signal
developed at the output of amplifier 75, as will be
discussed. A third feedback network includes p~rallel
coupled diodes 83 and 84, a feedback resistor 79, and
capacitor 77. Diodes 83 and 84 are mutually arranged to
conduct in response to opposite polarities of an applied
signal, as will also be discussed. An output signal (SO)
from processor 50 is coupled via a coupling capacitor 140
to the second input of combiner 30 (FIGURE 1).
Neglecting capacitors 90, 91 and 92 for the
moment, signal processor circuit 50 manifests a non-linear
composite amplitude transfer function as shown in FIGURE 5,
for imparting different amounts of signal gain to signals
having amplitudes within three ranges designated as I, II
and III in FIGURE 5. The value of capacitor 77 is chosen
sufficiently large so that the D.C. voltage across
capacitor ?7 substantially equals the D.C. level at the
- output of amplifier 75, after an initial settling time.
As mentioned above and as discussed in~
United States Patent No. 4,223,339Of W. A. Lagoni and
J. S. Fuhrer noted previously, it is important that the
detail information associated with small amplitude signals
be restored to prevent "smearing" of the vertical detail
content of the luminance signal. This function is provided
by resistor 87 in the circuit of FIGURE 2.
Diodes 81 and 82 are non-conductive for small
signal amplitudes. Resistor 87 is a linear device which,
in circuit 50, permits small amplitude detail signals to
be processed with the prescribed restoration gain (see
FIGURE 4) in region I, as shown by the amplitude transfer
characteristic of FIGURE 5. The restoration signal gain AI
(approximately two) imparted to output signal SO after
processing in region I is given by the following expression,
where R73, R78 and R87 correspond to the respective values
of resistors 73, 78 and 87, and Rp3 corresponds to the
~3 .
. . , . . i
28
1 - :L1 - RCA 73,993
parallel combination oE resistor 76 wit~ the series sum
of resistors 78 and 87:
AI = R x R- 78R
Resis-tor 87 does not affect the signal gain imparted to
moderate and large amplitude signals subject to processing
in regions II and III as will be discussed, since resistor 87
is "short circuited" or bypassed when diodes 81 and 82
conduct in response to applied moderate and large amplitude
signals.
As the amplitude of the signal appearing at the
15 output of amplifier 75 increases above the threshold
conduction level of diodes 81 and 82 by a moderate amount
(correspondin~ to signals subject to processing in
region II), diodes 8]. and 82 are caused to conduct. The
signal gai.n for processor circuit S0 in region II (AII) is
20 then determined to be greater than the restoration gain,
or approximately three, in accordance with the following
expression, where Rpl corresponds to the value of the
parallel combination of resistors 76 and 78, and R73
corresponds to the value of resistor 73:
A RPl
II R73
The gain imparted to moderate amplitude signals is
illustrated by the amplitude transfer function for region II
30 in FIGURE 5. In this;connection it is no-ted that small
amplitude excursions of moderate amplitude signals are
translated with the restoration gain, while the peak
amplitude excursions are amplified as indicated above. The
width of region I for both signal polarities is a function
35 of the ratio of the value of resistor 76 to the value of
resistor 73.
The amplitude transfer function associated with
large amplitude detail signals corresponding to those
signals associated with region III in FIGURE 5, is
4~
-
: ' '
.
~3~28
1 ~ 12 - RCA 73,993
determined by the feedback network includiny resis-tor 79
and paring diodes ~3 and 8~, which are rendered conduc-tive
in response to the peak amplitude excursions of the large
amplitude signals. Diodes 81 and 82 also conduct a-t -this
time. This network serves to pare the peak amplitude
excursions of the large amplitude detail signals in
region III by causing -these peak excursions to be processed
with less than the restoration galn. The signal gain in
this instance for region III (AI[I) is given by the
following expression, where Rp2 corresponds to the value
of the parallel combination of resistors 76, 78 and 79,
and R73 corresponds to the value of resistor 73:
R
P2
III R73
In this connection it is noted that small amplitude
excursions of the large amplitude signals are translated
with the restoration gain, moderate amplitude excursions
are processed with greater -than the restoration gain, and
peak amplitude excursions are processed with less than the
restoration gain.
Thus the composite amplitude transfer fucntion of
circuit 50, as shown by FIGURE 5, exhibits three gain
regions for three predetermined levels of signal amplitude,
for both positive (-~) and negative (-) signal polarities.
The frequency response of signals processed within any of
the three gain regions can be tailored by employing filter
capacitors such as capacitors 90, 91 and 92 in parallel
with the appropriate feedback resistor as shown. In this
example, the frequency bandwidth of processed small
amplitude signals (region I) is proportional to the
reciprocal of the value of capacitor 90. The frequency
35 bandwidth of processed moderate amplitude signals
(region II) is proportional to the reciprocal of the sum
of the values of capacitors 90 and 91. The frequency
bandwidth of processed large amplitude signals (region III)
is proportional to the reciprocal of the sum of the values
3~8
1 - 13 - RCA 73,993
of capacitors 90, 91 and 92.
The circuit of FIGURE 3 ls iden-tical to -that of
FIGURE 2, except -that resistor 87 in FIGURE 2 has been
removed. The amplitude transfer function for the circuit
of FIGURE 3 is given by FIGURE 6. This transfer function
differs from that of FIGURE 5 in that small amplitude
signals are "cored", or inhibited, in region I due to
diodes 81 and 82 being non-conductive at this time. This
response corresponds to a zero gain response in region I,
whereby small amplitude signals do not appear as output
signals (SO)- Small amplitude signals do no-t appear as
output signals (SO) since diodes 81, 82 and 83, 84 are
non-conductive at this time and the inverting input (-)
of operational amplifier 75 represents a virtual ground
point, causing no signal currents to flow to the circuit
output via resistors 78, 79 or capacitors 77, 91, 92 under
these conditions.
FIGURE 7 illustrates another circuit embodiment
of processor 50, which is similar to the circuit arrangement
of FIGURE 2. Corresponding elements in the circuits of
FIGURES 2 and 7 are identified by the same reference
numbers.
In this embodiment, a common emitter amplifier
transistor 75 has a base input electrode corresponding to
an inverting input, a grounded emitter electrode, and a
collector output electrode coupled to a source of operating
supply voltage (+16 volts) via a collector load impedance
112. The open loop g~in for transistor 75 is primarily
determined by the value of load impedance 112, and is
sufficiently high to approximate the open loop gain of an
operational amplifier (e.g., amplifier 75 in FIGURE 2).
Vertical detail signals (Si) from filter 35 are
supplied to the base input of transistor 75 via a network
including input resistor 73, a capacitor 105 and a
resistor 106 arranged as shown. The latter two elements
are employed to frequency compensate the open loop
frequency response of transistor 75. The input signal (Si)
- , - ~
-- . .
- RCA 73,993
need not be A.C. coupled when the input si~nal exhibits a
stable, predictablc ~.C. levc~l as is assumed to be the
case in this example, in which the input signal is D.C.
coupled to the base of transistor 75. This D~C. ].evel in
conjunction with resistors 73, 76 and 114 serve to
establish a desired operatinq point for transistor 75.
As in -the case of the FIGURE 2 circuit, resistor
76 in FIGURE 7 determines the width of region I (i.e., the
restoration gain region for both signal polarities) in
conjunction with resistor 73 and the threshold conduction
level of diodes 81, 82. Feedback capacitor 77 exhibits
low D.C. current leakage and assists to maintain diodes
81-84 properly biased to produce a symmetrical transfer
function. The output signal (SO) is A.C. coupled via a
capacitor 140 to inhibit D.C. current flow which can upset
the desired symmetry of the composite transfer function
(FIGURE 5). The frequency response of signals processed
within any of the three gain regions can be tailored by
employing capacitive feedback as explained in connection
with FIGURE 2.
For either of the ci~cuit arrangements of
FIGURES 2 or 7, the composite transfer function shown by
FIGURE 5 can be modified to suit the requirements of a
given system. For example, the response in region I can
be eliminated by removing resistor 78 and short circuiting
diodes 81, 82 and resistor 87 (e.g., by connecting a wire
across resistor 87). Low level signal restoration in
region I can be eliminated by removing resistor 87, as
indicated by FIGURES 3 and 6. Large amplitude signal
paring in region III can be eliminated by removing diodes
83, 84 and resistor 79.
It is also noted that the described vertical
detail signal processing arrangement is unaffected by
variations in the D.C. level of the luminance component.
Due to the manner in which a comb filter derives a combed
chrominance signal by employing a subtractive signal
combining process as is known, the combed chrominance
1 - 15 - RC~ 73,993
signal exhibits a zero D.C. component. The D.C. componen-t
of the combed chrominance signal therefore does not upset
5 the D.C. bias component developed at the comb filter
chrominance output. Processing of the vertical detail
signal, as derived from the comb fil-ter chrominance output,
can therefore be centered predictably about -the D.C. bias
component when the comh filter chrominance output is D.C.
10 coupled to the vertical detail signal processing network.
Since the reference level about which signal processing is
accomplished is fixed predictably, well-defined restoration,
enhancement and paring regions result.
~0
.
. - . . ~ .
' ' ' ' ' ' ' - '