Digital encoding of color video signals

• US 4,141,034 A
• Filed: 08/19/1977
• Issued: 02/20/1979
• Est. Priority Date: 08/19/1977
• Status: Expired due to Term

• Current Assignee: Bell Telephone Laboratories, Inc. (Nokia Corporation)
• Original Assignee: Bell Telephone Laboratories, Inc. (Nokia Corporation)
• Inventors: Netravali, Arun N., Rubinstein, Charles B.
• Primary Examiner(s): Griffin, Robert L.
• Assistant Examiner(s): Psitos, Aristotelis M.
• Application Number:
  US05/826,025
• Time in Patent Office: 550 Days
• Field of Search: 358/13, 358/135, 358/140, 325/38 B
• US Class Current: 375/240.23
• CPC Class Codes: G06T 9/004 Predictors, e.g. intraframe, interframe coding
• CPC Class Codes: H04N 11/046 DPCM

Abstract

For a given transmission bit rate, the efficiency of encoding the color components of a color video signal is increased by utilizing information obtained from the luminance component to optimize color component encoding. Specifically, several predictions of the present luminance sample are formed using various individual values or combinations of previous luminance information. Each prediction is evaluated, and the best prediction technique is used to predict the present color sample, based upon previous color component information. Decoding in a remote receiver is accomplished similarly, by utilizing the luminance signal as a source for color component prediction information.

Claims

1. Color video signal encoding apparatus for predicting the chrominance value representing the color information for a picture element in said signal, said picture element having a luminance value L_n, comprising:

(a) means for forming a first prediction of said luminance value of said picture element based upon the luminance value of a first previous picture element, and a second prediction of said luminance value based upon the luminance value of a second previous picture element;

(b) means for computing the differences between each of said first and second predictions and said luminance value L_n ; and

(c) means responsive to said computing means for predicting said chrominance value based upon the chrominance value of either said first or said second previous picture element, depending upon which of said luminance predictions produces a smaller difference.

2. The invention defined in claim 1 further including:

(a) a quantizer for generating a fixed length output word having a value representative of the difference between said predicted chrominance value and the true value thereof, and

(b) a variable length encoder responsive to said quantizer output for generating an output code having a length which varies as a function of the frequency of occurrence of said quantizer output value.

3. The invention defined in claim 2 further including:

(a) means for comparing the smaller of said differences between each of said first and second predictions and said luminance value L_n with a threshold value; and

(b) means for varying the length of said output code as a function of the output of said comparing means.

4. Color video signal encoding apparatus for predicting the chrominance value of a picture element (pel) in the picture represented by said signal, said pel having a true luminance value, comprising:

(a) means for forming a plurality of predictions of said true luminance value of said pel by combining the luminance values of neighboring pels in a corresponding plurality of different patterns;

(b) means for comparing each of said predicted luminance values to said true luminance value to select the one of said patterns producing the smallest luminance prediction error; and

(c) means for combining the chrominance values of said neighboring pels in accordance with said selected pattern.

5. The invention defined in claim 4 further including:

(a) means for computing the difference between said predicted chrominance value and the true value thereof; and

(b) means for encoding said difference as a code having a length which varies as a function of the value of said smallest luminance prediction error.

6. In a DPCM encoder for encoding the color component of a video signal which has a corresponding luminance component, apparatus for predicting the color value of a particular picture element based upon the color values of neighboring picture elements, comprising:

(a) means for forming at least two predictions of the luminance value of said particular picture element by computing at least two different weighted sums of the luminance values of said neighboring picture elements;

(b) means for selecting from among said two different weighted sums an optimum weighted sum which produces the smallest luminance prediction error; and

(c) means responsive to said selecting means for computing the optimum weighted sum of color values of neighboring picture elements.

7. Apparatus for encoding a color video signal including a luminance component and at least one color component, comprising:

(a) first means for storing past samples of said luminance component;

(b) second means for storing past samples of said color component;

(c) third means responsive to said first means for forming a plurality of predictions of the present value of said luminance component using various combinations of said past luminance samples;

(d) fourth means for evaluating each of said predictions to determine a particular one of said combinations which best approximates said present luminance value; and

(e) fifth means responsive to said second and fourth means for forming a prediction of the present value of said color component by combining said past color samples using said particular one of said combinations.

8. The invention defined in claim 7 further including a DPCM encoder arranged to generate an output representing the difference between said present and predicted color values;

and

a variable length coder for generating a code word of variable length jointly in response to (1) said output of said DPCM encoder and (2) the difference between said present luminance value and said particular combination.

9. Apparatus for encoding a color video signal including a luminance component and at least one chrominance component, comprising:

(a) means for sampling said luminance and chrominance components at a plurality of picture elements;

(b) means for forming at least two predictions of the luminance value of one of said picture elements based upon the luminance values of at least first and second different ones of said picture element;

(c) means for evaluating said first and second predictions to determine which of said different picture elements provide the best luminance prediction; and

(d) means for forming a prediction of the chrominance value of said one picture element based upon the color value of said different picture elements which provide said best luminance prediction.

10. The invention defined in claim 9 wherein said prediction forming means includes a delay circuit.

11. The invention defined in claim 10 wherein said sampling means is arranged to sample said signal along a plurality of parallel scan lines, and

said delay circuit includes a first output having a one-sample delay and a second output having a one-line delay.

12. A method of encoding the color component of a video signal which has a corresponding luminance component, including the steps of:

(a) combining past samples of said luminance component in at least two different patterns to generate at least two predictions of the present value of said luminance component;

(b) subtracting each of said predictions from said present value to determine a particular one of said combination patterns which produces the smallest prediction error; and

(c) combining past samples of said color component in said particular pattern to generate a prediction of the present value of said color component.

13. The method of claim 12 further including the steps of:

(d) quantizing the difference between said present color value and said prediction thereof; and

(e) encoding said quantized difference as a code word having a length which varies as a function of the magnitude of said difference and said smallest prediction error.