Apparatus for combining and separating constituent components of a video signal

<td> 1*1</td><td> Consommation et Affaires commerciales Canada</td><td colspan="4"> Consumer and Corporate Affairs Canada</td>

<td></td><td> Bureau des brevets</td><td> Patent Office</td><td></td><td></td><td></td>

<td></td><td> Ottawa, Canada K1A 0C9</td><td></td><td> (11)</td><td> (C)</td><td> 1,326,537</td>

<td></td><td></td><td></td><td> (21)</td><td></td><td> 608,647</td>

<td></td><td></td><td></td><td> (22)</td><td></td><td> 1989/08/17</td>

<td></td><td></td><td></td><td> (45)</td><td></td><td> 1994/01/25</td>

<td></td><td></td><td></td><td> (52) C.L.</td><td> CR.</td><td> 350-40 350-33 350-56</td>

(51) INTL.CL. H04N—011/00; H04N-009/78; H04N-007/13 (19) (CA) CANADIAN PATENT (12) (54) Apparatus for Combining and Separating Constituent Components of a Video Signal (72) Isnardi, Michael A. , U.S.A.

(73) General Electric Company , U.S.A.

(30) (US) U.S.A. 253273 1988/10/03 (57) 30 Claims

Canada

CCA 3254 (10-92) 41 7530-21-935-3254 r’ .

<img file="CA1326537C_D0001.tif" />

RCA 84,680

The luminance component of a video signal is processed to produce a luminance component with groups of mutually exclusive spatially correlated image pixels of 5 identical value within each group. In one embodiment each group comprises vertically averaged pixels 1H apart within a field, In another embodiment each group comprises vertically averaged pixels 263H apart within a frame. The chrominance component i’s similarly processed. The processed luminance and chrominance components are combined using an invertible algorithm involving additive and subtractive signal combination. At a decoder, the inverse of the combining algorithm is used to separate the combined signal into its constituent luminance and chrominance component parts, without luminance-chrominance crosstalk.

♦

-1- 1326537

RCA 84,680

APPARATUS FOR COMBINING AND SEPARATING CONSTITUENT COMPONENTS OF A VIDEO SIGNAL

This invention concerns apparatus for encoding video signal components to be combined, such as the luminance and chrominance components of an NTSC-type television signal, to facilitate the separation of such components without crosstalk artifacts.

It is well-known that the frequency interleaving of the baseband luminance and chrominance components of an NTSC-type television signal does not always permit the two components to be separated without crosstalk artifacts such as cross-color and cross-luminance artifacts. The cross-color artifact results from chrominance contamination of luminance information, and can be described as a scintillating rainbow pattern in parts of an image where diagonal high frequency luminance information is present, such as on a striped shirt. The cross-luminance artifact is sometimes referred to as the hanging dot artifact, and is visible in a system using a comb filter type luminance/chrominance separator. Cross-luminance results from luminance contamination of chrominance information particularly at vertical image transition regions.

Cross-color and cross-luminance artifacts are introduced at two places in the system. First, imperfect or incorrect chrominance/luminance separation at a television receiver causes luminance to be interpreted as chrominance and vice-versa. The second way in which cross-color and cross-luminance artifacts are introduced is in the encoding process itself, which frequency interleaves the luminance and chrominance components.

Various methods have been proposed to improve the separation of luminance and chrominance components at the receiver by reducing luminance/chrominance crosstalk artifacts. Some of these methods involve source and receiver processing schemes using line, frame and field comb filters, such as described by C. H. Strolle in

<img file="CA1326537C_D0002.tif" />

-2132053F^{A 84}'⁶⁸⁰

Cooperative Processing of Improved NTSC Chrominance/ Luminance Separation, SMPTE Journal, August 1986. Another system for prefiltering luminance and chrominance information via multi-dimensional comb filters before being combined is disclosed by Faroudja et al. in Improving NTSC to Achieve Near-RGB Performance, SMPTE Journal, August 1987.

The previously proposed methods which seek to eliminate luminance/chrominance crosstalk artifacts suffer from one or more practical drawbacks related to switching artifacts in motion adaptive processing systems, or to filtering complexity in terms of memory requirements and the number of filter taps, for example. The luminance/chrominance separation system disclosed herein in accordance with the present invention uses a cooperative processing technique with matched encoding and decoding processes. The herein disclosed system exhibits enhanced luminance/chrominance separation in the vertical and vertical-temporal directions with minimal image degradation, and avoids many of the drawbacks of the known systems. In addition, the disclosed system is compatible with the existing NTSC system.

In accordance with principles of the present invention, at a video signal encoder a signal component is processed to produce a first component with groups of mutually exclusive image pixels of identical value within each group and separated by an odd number of image scanning lines. Similarly, a further signal component is processed to produce a second component with groups of mutually exclusive image pixels of identical value within each group and separated by an odd number of image scanning lines. The processed first and second components are combined by means of an invertible algorithm. At a decoder, the inverse of the combining algorithm is used to separate the combined signal into its constituent

-3RCA 84,680 processed first component and processed second component parts, without component crosstalk.

In an illustrated embodiment of the invention, each group comprises a pair of vertically averaged spatially correlated pixels one horizontal scanning line (1H) apart, with luminance information being processed only for frequencies above the vertical detail band.

The invertible combining algorithm involves additive and subtractive combination of a luminance component with a chrominance component modulating a chrominance subcarrier. In another illustrated embodiment of the invention, each group comprises a pair of averaged spatially correlated pixels 263H apart.

Description of the Drawing

FIGURE 1 is an illustration of a luminance/chrominance encoding and decoding technique according to an embodiment of the present invention, using a 1H line interval encoding process;

FIGURE 2 is a block diagram of apparatus implementing the technique of FIGURE 1;

FIGURES 3 and 4 show aspects of the apparatus of FIGURE 2 in greater detail;

FIGURE 5 is an illustration of a luminance/chrominance encoding and decoding technique according to a further embodiment of the present invention, using a 263H field interval encoding technique;

FIGURE 6 is a block diagram of apparatus implementing the technique of FIGURE 5;

FIGURES 7 and 8 show aspects of the apparatus of FIGURE 6 in greater detail; and

FIGURE 9 shows a portion of the apparatus of FIGURE 2 in greater detail.

FIGURE 1 illustrates one process by which luminance and chrominance components of a standard NTSC television signal are encoded at a transmitter and decoded at a receiver. In FIGURE 1, an odd image field before encoding is shown as having interlaced baseband luminance

-41326537

RCA 84,680 (Y) information associated with odd horizontal image scanning lines, designated as Yl, Y3, Y5, Y7 etc., and interlaced baseband chrominance (C) information associated with odd horizontal image scanning lines, designated as Cl, C3, C5, C7 etc.. Within each odd field (and even fields not shown) of a given frame, mutually exclusive non-overlapping groups of spatially correlated picture element (pixel) pairs 1H apart are processed by averaging line-pairs vertically and replacing the original pixel values with the average values within each pixel group. This process is performed separately for luminance and chrominance information. The averaging process is performed for all chrominance horizontal frequencies, but is performed only for luminance horizontal frequencies above about 1.8 MHz to preserve vertical detail information below about 1.8 MHz. The averaged chrominance information modulates a 3.58 MHz chrominance subcarrier haying an opposite phase from line-to-line within a field before being combined with the averaged luminance information in a standard manner to form a composite baseband color video signal. The modulated chrominance signal is then combined with the processed luminance information. The line-to-line opposite phase exhibited by the chrominance information as a result of the subcarrier modulation process produces additive and subtractive combinations of luminance and chrominance information within each group of combined luminance/chrominance information. The additive and subtractive luminance/ chrominance combinations associated with the combination process (e.g., Y₁₊₃ ± C₁₊₃) represent an invertible algorithm to facilitate the separation of the luminance and chrominance information at the receiver.

More specifically, the phase of the chrominance subcarrier changes so that the phase of the modulated chrominance signal is the same from line-to-line for samples 262H apart within an image frame, but is opposite for odd numbered line samples 1H apart within a field and 263H apart within a frame. This line-to-line phase change provides the

-5RCA 84,680 subtractive aspect of the additive-subtractive invertible algorithm.

Decoding at a receiver employs a line-pair decoding process within a field. A received encoded composite signal is separated into low and high horizontal frequency bands. Above about 2 MHz, the mutually exclusive non-overlapping groups containing line-pairs 1H apart are summed to produce high frequency luminance information, and are differenced to yield the modulated chrominance information. The high frequency luminance information is added to the low frequency luminance information to reconstitute the luminance signal, and the modulated chrominance information is demodulated in the standard manner. The 1H line-pair processing system of FIGURE 1 exhibits enhanced luminance-chrominance separation in the vertical dimension. The luminance signal produced after line-pair decoding (Y₁₊₃,

Y₅₊₇ etc.) does not contain cross-color artifacts, and the chrominance component produced after line-pair decoding (^c^₊₃, ^C5+7 ^etc·) lacks cross-luminance artifacts. The line-pair averaging used in the encoding process has a negligible effect on image resolution because of the high correlation of image information a line apart within a field, or a field apart within a frame.

The encoding/decoding process of FIGURE 1 can be implemented as shown in FIGURE 2, as follows. In a transmitter encoder, a source of color T.V. signals 10, e.g., including a color camera, provides a luminance component Y and chrominance color difference signal components I and Q in digital form. The luminance component is filtered by means of a 1.8 MHz horizontal lowpass filter 12 to produce a low frequency luminance signal YL. Signal YL, after inversion by an inverter 14, is combined in an adder 16 with the original luminance signal (Y) to produce a high frequency luminance signal YH. A delay matching unit 18 equalizes the transit times of the signals combined in adder 16 by compensating for the delay of filter 12. Signal YH, containing luminance frequencies

-61326537

RCA 84,680 above approxiamtely 1.8 MHz, is subjected to a 1H line pair averaging process within a field, as discussed in connection with FIGURE 1, by means of a processor 20. A signal YH* from processor 20 is combined in an adder 32 with low frequency luminance signal YL’ from the output of a 1H line delay network 21.

The I and Q color difference signal components from source 10 are filtered by a 0.5 MHz lowpass filter 22 and a 1.3 MHz lowpass filter 24, respectively, before being separately subjected to a 1H line-pair averaging process in units 26 and 28. Line-pair averaged chrominance signals 1' and Q' from units 26 and 28 are applied to a standard NTSC modulator 30 for quadrature modulating a 3.58 MHz chrominance subcarrier SC. Included in unit 30 is a delay matching network for equalizing the transit times of the color difference signals from units 26 and 28 before modulating the chrominance subcarrier. A modulated chrominance signal C’ from modulator 30 exhibits line-to-line phase reversal and is combined in an adder 33 with signal Y' from adder 32 to produce an output NTSC signal which is conveyed via suitable transmission apparatus, e.g., including RF modulator networks, to a television signal receiver. A line identifier signal F^ is provided by a line identifier network 11 to switching control inputs of units 20, 26 and 28. Network 11 responds to luminance component Y, particularly to the synchronizing information of luminance component Y. Signal F_T, together with a similar signal F_T ’ provided to receiver network 48 by a receiver line identifier network 41 in response to the received NTSC signal, assures that the line encoding and decoding processes are synchronized to occur with respect to the same line pairs. Apparatus suitable for use as networks 11 and 41 is shown in FIGURE 9.

At the television signal receiver, after RF demodulation and associated conventional input processing to produce a baseband signal (not shown), the NTSC signal is filtered by means of a 2 MHz lowpass filter 40 to produce low frequency luminance signal YL*. The 2 MHz

-71326537

RCA 84,680 cut-off frequency of filter 40 together with the 1.8 MHz cut-off frequency of encoder filter 12 produces a 200 KHz guardband to reduce horizontal crosstalk. Signal YL·', after inversion by an inverter 42, is combined with the received full bandwidth NTSC signal in a combiner 44 to produce a high frequency video signal above 2 MHz. A delay network 46 compensates for the delay exhibited by filter 40 and equalizes the transit times of the signals combined by unit 44. The high frequency signal from combiner 44 is processed by an averaging and differencing unit 48 which processes line-pairs 1H apart to separate the input signal into high frequency luminance and chrominance components YH’ and C. High frequency luminance signal YH' is combined in an adder 52 with low frequency luminance signal YL* after signal YL* is subjected to a 1H line delay via a unit 50, for producing a full bandwidth luminance signal Y’ .

Modulated chrominance component C from unit 48 is conveyed to a unit 54 which provides delay matching and quadrature demodulation functions to produce color difference signals I’ and Q’. Signals 1' and Q' are processed by a chrominance processor 56 including gain control and phase shifting circuits for example, and luminance signal Y’ is processed by a luminance processor 58 illustratively including gain control and level shifting circuits, before being applied to a matrix 60 for combining the luminance and chrominance signals to produce red, green and blue image representative signals R, G and B. A display driver stage 62 amplifies these signals to a level suitable for driving an image display kinescope 64.

FIGURE 3 shows a line-pair averager circuit suitable for use in units 20, 26 and 28 of the encoder arrangement in FIGURE 2. Input signals are filtered by a vertical lowpass filter 70 which is included to reduce vertical aliasing artifacts that may occur with certain diagonal scene material. Filter 70 is optional. The filtered input signal is coupled via a 1H line delay unit 72 to an input A of an averaging network 74, and is coupled

-a1326537

RCA 84,680 directly to an input B of averaging network 74. A line averaged signal from network 74 is coupled directly to one input of a line rate multiplexer (MUX) 76, and to another input of MUX 76 via a 1H line delay 78. Signal F_T from

JU network 11 is applied to a switching control input of MUX 76.

FIGURE 4 shows details of line-pair averager/ differencer 48 in FIGURE 2. A unit 80 receives a line delayed version of the input signal at an input A via a 1H line delay unit 82, and receives the input signal itself at an input B. At one output an averaged signal (A+B)/2 is produced, and a differenced signal (A-B)/2 is produced at another output. 1H line delayed and undelayed versions of the averaged signal from unit 80 are conveyed via a line rate MUX 83 and a vertical lowpass filter 84 to an output at which signal YH is produced. 1H line delayed and undelayed versions of the differenced signal from unit 80 are conveyed via a line rate MUX 86 and a vertical bandpass filter 88 to an output at which chrominance signal C is developed. Signal F^ is applied to a switching control input of MUX 83 and MUX 86. Output filters 84 and 88 are optional and serve essentially the same purpose as filter 70 in FIGURE 3.

FIGURE 5 illustrates another method by which luminance and chrominance components of a standard NTSC television signal are encoded at a transmitter and decoded at a receiver without luminance/chrominance crosstalk. The method illustrated by FIGURE 5 is similar to that of FIGURE 1 except that in FIGURE 5 263H field pair processing rather than 1H line-pair processing is employed.

In FIGURE 5 each odd field includes interlaced baseband luminance information associated with odd horizontal image scanning lines Yl, Y3 etc., and associated interlaced baseband chrominance information Cl, C3 etc.. Each even field includes interlaced baseband luminance information associated with even horizontal image scanning lines Y2, Y4 etc., and associated interlaced baseband chrominance information C2, C4 etc.. Within each frame,

-9RCA 84,680 line pairs 263H apart are processed such that mutually exclusive non-overlapping groups of pixels 263H apart are averaged vertically, with the average values replacing the original values within each group. This process is performed separately for luminance and chrominance information. The averaging process is performed for all chrominance horizontal frequencies, but is performed only for luminance horizontal frequencies above about 1.8 MHz to preserve luminance vertical detail information below about 1.8 MHz. The averaged chrominance information modulates a 3.58 MHz chrominance subcarrier having an opposite phase from line-to-line, before being combined with the averaged luminance information in a standard manner to form a composite baseband color video signal. The line-to-line phase inversion exhibited by the modulated chrominance signal as a result of the chrominance modulation process produces additive and subtractive combinations, of luminance and chrominance information within each group of combined luminance/chrominance information. As in the case of the FIGURE 1 system, the additive and subtractive luminance/chrominance combination represents an invertible algorithm to facilitate luminance/chrominance separation at the receiver.

Decoding at a receiver is essentially the same as discussed in connection with the arrangement of FIGURE 1, except that the arrangement of FIGURE 5 employs a 263H field pair decoding process within a frame. The field pair processing system of FIGURE 5 exhibits enhanced luminance-chrominance separation in the vertical-temporal dimension without cross-color or cross-luminance artifacts.

The encoding/decoding process of FIGURE 5 can be implemented as shown in FIGURE 6. The arrangement of FIGURE 6 is similar to that of FIGURE 2 except that FIGURE 6 uses 263H field processing networks 120, 121, 126, 128, 148 and 150 rather than 1H line processing networks, and field rather than line identifier networks for producing field identifier signals F_ and F' .

-101326537

RCA 84,680

FIGURE 7 shows a 263H field-pair averager network suitable for use in encoder field pair averaging networks 120, 126 and 128 of FIGURE 6. FIGURE 8 shows a field-pair averager/differencer suitable for use in decoder network 148 of FIGURE 6. The apparatus of FIGURES 7 and 8 are similar to the apparatus of FIGURES 3 and 4 except that the apparatus of FIGURES 7 and 8 employ 263H field-pair processing units rather than 1H line-pair processing units.

FIGURE 9 shows apparatus suitable for use as networks 11 and 41 of FIGURE 2. An input signal, such as luminance component Y in the case of generator 11, is applied to a sync separator and timing signal generator 160. Generator 160 includes synchronizing signal separator and sensing circuits and counter circuits for producing a horizontal blanking interval signal which is applied to a clock (CLK) input of a line counter 165, and for producing a vertical blanking interval signal which is applied to a RESET input of line counter 165. Counter 165 is reset at the end of each vertical blanking input signal and then initiates a horizontal line counting process in response to the clock input horizontal blanking signal. The OUTPUT signal is obtained by sensing the least significant bit of the line count to produce a low logic level for even lines and a high logic level for odd lines, for example.

Apparatus suitable for use as the field identifier networks shown in FIGURE 6 for producing field identifier signals F_Fand F’_f can be readily developed as explained, for example, in the text Circuit Concepts by Gerald A. Eastman, pp. 88-92, available from Tektronix, Inc., Beaverton, Oregon.

The principles of the present invention can be used to encode, for combining and subsequent separation, signals other than luminance and chrominance components as disclosed. In such case one of the signals to be combined should exhibit a line-to-line phase change (i.e., having substantial negative correlation) within a field or within a frame so as to develop an invertible algorithm such as the illustrated algorithm employing additive and subtractive signal combination. The required phase change

1326537_{RCA θ4},_6β0 may be inherent in the nature of such signal, or it may be produced by external means such as a modulation process.

The principles of the present invention also can be used to develop a cooperative frame processor encoder/ 5 decoder arrangement where elements to be grouped are 525H apart, i.e., a frame apart, rather than 1H or 263H apart as illustrated.

-12- 1326537 RCA 84,680