Coding technique for high definition television signals

• US 5,337,085 A
• Filed: 04/10/1992
• Issued: 08/09/1994
• Est. Priority Date: 04/10/1992
• Status: Expired due to Fees

• Current Assignee: COMSAT Corporation (Reorganized ISA SA)
• Original Assignee: COMSAT Corporation (Reorganized ISA SA)
• Inventors: Bhargava, Sanjai, Lee, Lin-Nan, Rao, Ashok K.
• Primary Examiner(s): Britton, Howard W.
• Application Number:
  US07/866,851
• Time in Patent Office: 851 Days
• Field of Search: 358/133, 358/12, 358/13, 358/135, 358/136, 375/27, 348/398, 348/402, 348/410, 348/414, 348/416, 348/417
• US Class Current: 375/240.11
• CPC Class Codes: G06T 9/008 Vector quantisation
• CPC Class Codes: H04N 19/30 using hierarchical techniques, e.g. scalability H04N19/63 takes precedence
• CPC Class Codes: H04N 19/94 Vector quantisation

Abstract

A coding technique for high-definition television signals involves splitting the luminance signal into four sub-bands using a sub-band encoder and separately encoding each sub-band using a simplified vector quantization technique. The characteristics of each respective simplified vector quantization encoder can be matched to the characteristics of the respective sub-bands.

Claims

1. A television signal coding method comprising the steps of:

(a) dividing an image into a plurality of two-dimensional blocks, each block being composed of a predetermined number of pixels;

(b) for each block exhibiting an intensity variation within that block, calculating first and second means values based on said variation;

(c) calculating distortion values between said each block and a respective plurality of basic vectors in a codebook based on said first and second mean values, said vectors corresponding to the most common patterns to which the human visual system is highly sensitive;

(d) identifying one of said vectors producing a minimum one of said distortion values as a selected pattern; and

(e) generating a digital signal including a plurality of first bits representing said selected pattern, a plurality of second bits representing said first mean value and a plurality of third bits representing step size of a pattern edge in said selected pattern.

2. The method according to claim 1, wherein said predetermined number of pixels is adaptively varied inversely proportional to magnitude of said minimum difference.

3. The method according to claim 1, wherein each vector comprises n bits and there are less than 2ⁿ vectors in said codebook.

4. A television signal coding method comprising the steps of:

(a) dividing a video signal into its luminance and chrominance components;

(b) dividing the luminance component into a plurality of sub-bands;

(c) simultaneously performing the following coding on each sub-band;

(i) dividing an image into a plurality of two-dimensional blocks, each block being composed of a predetermined number of pixels;

(ii) for each block exhibiting an intensity variation within said block, determining a minimum difference between said block and codebook patterns permitting selection of one of said codebook patterns as a selected pattern using at least one mean value representing a corresponding each said block; and

(iii) generating a digital signal including a plurality of first bits representing said selected pattern, a plurality of second bits representing said means value and a plurality of third bits representing step size of a pattern edge in said selected pattern;

(d) simultaneously coding said chrominance component;

(e) combining the thus-coded luminance and chrominance components into a single bit stream for transmission.

5. A method according to claim 4, wherein said predetermined number of pixels is adaptively varied inversely proportional to magnitudes of said distortion values.

6. A method according to claim 4, wherein said means value comprises first and second mean values and wherein said determining step (c)(ii) comprises the steps of:

(c)(ii)(1) calculating said first and second mean values based on said variation;

(c)(ii)(2) calculating distortion values between each said block and each of said codebook patterns based on said first and second mean values;

(c)(ii)(3) identifying the one of said codebook patterns producing a minimum distortion value corresponding to said minimum difference.

7. A method according to claim 4, wherein said determining step (c)(ii) comprises the steps of:

(c)(ii)(1) calculating mean intensity for each block and subtracting the mean intensity from every pixel intensity of the block;

(c)(ii)(2) replacing positive and negative values resulting from step (c)(ii)(1) with zeros and ones, respectively, to generate a resultant block representing said each block;

(c)(ii)(3) constructing a digital word by scanning said resultant block from left to right and top to bottom; and

(c)(ii)(4) entering said digital word in a look-up-table relating Hamming distance between said digital word and said codebook patterns so as to identify said selected pattern.

8. A method according to claim 4, wherein said television signal is a High Definition Television signal and wherein one of said sub-bands represents a NTSC luminance signal.

9. A method according to claim 4, wherein said step (b) further comprises dividing the luminance component into a first sub-band and a plurality of sub-bands based on frequency;

and wherein said step (c) further comprises simultaneously performing interframe coding of said first sub-band and intra frame coding on each of said second sub-bands.

10. The method according to claim 4, wherein said coding recited in step (c) is interframe coding.

11. A method according to claim 4, wherein said plurality of vectors in said codebook comprise only basic vectors corresponding to the most common patterns to which the human visual system is highly sensitive, wherein each vector comprises n bits and there are less than 2ⁿ vectors in said codebook.

12. A television signal coding method for compressing said television signal using a codebook comprising only two mean edge patterns to which human sight is highly sensitive, said method comprising the steps of:

(a) dividing an image into a plurality of two-dimensional blocks, each block being composed of a predetermined number of pixels;

(b) calculating mean intensity for each block and subtracting the mean intensity from every pixel intensity of the block;

(c) replacing positive and negative values resulting from step (b) with zeros and ones, respectively, to generate a resultant block representing said each block;

(d) constructing a digital word by scanning said resultant block from left to right and top to bottom; and

(e) entering said digital word in a look-up-table relating a difference between said digital word and said patterns so as to identify a selected pattern; and

(f) generating a digital codeword based on said selected pattern.

13. The method according to claim 12, wherein said difference corresponds to Hamming distance.

14. The method according to claim 12, wherein said digital codeword comprises a plurality of first bits corresponding to said selected pattern, a plurality of second bits representing at least one of mean intensity values in said block and a plurality of third bits representing edge step size.

15. The method according to claim 12, wherein said predetermined number of pixels is equal to an integer N greater than 1, and wherein look-up-table is formed by the steps of:

(g) for each of said patterns in said codebook, generating one corresponding N-bit word;

(h) generating a list of N bit entry words corresponding to 2^N possible N-bit words; and

(h) correlating each of said N bit entry words with one of said patterns representing approximately minimum Hamming distance.

16. The method according to claim 15, wherein said patterns in said codebook are ordered based on frequency of occurrence.