Hierarchical encoding and decoding apparatus for a digital image signal

• US 5,598,214 A
• Filed: 09/28/1994
• Issued: 01/28/1997
• Est. Priority Date: 09/30/1993
• Status: Expired due to Term

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Kondo, Tetsujiro, Fujimori, Yasuhiro, Nishikata, Takeharu
• Primary Examiner(s): Britton, Howard W.
• Application Number:
  US08/313,811
• Time in Patent Office: 853 Days
• Field of Search: 348/409, 348/410, 348/411, 348/412, 348/414, 348/417, 348/418, 348/421, 348/424
• US Class Current: 375/240.14
• CPC Class Codes: H04N 19/102 characterised by the element, parameter or selection affected or controlled by the adaptive coding
• CPC Class Codes: H04N 19/61 in combination with predictive coding
• CPC Class Codes: H04N 19/63 using sub-band based transform, e.g. wavelets
• CPC Class Codes: H04N 19/645 by grouping of coefficients into blocks after the transform
• CPC Class Codes: H04N 19/94 Vector quantisation

Abstract

In a hierarchical encoding apparatus, predictors each including a look-up table are used to form predicted signals for respective hierarchical levels. The difference between the predicted signal and the original signal forms the differential signal for the higher hierarchical levels. The look-up table contains either predictive coefficients which are used as weights for pixels of a signal in one hierarchical level to produce a signal at a higher hierarchical level, or representative values which are combined with a dynamic range and redundant component in a set of pixels of a signal in one hierarchical level to produce a signal at a higher hierarchical level. The values in the look-up table are obtained from representative images.

Claims

1. An apparatus for encoding a first digital image signal representing pixels and having a first resolution, comprising:

means for forming said first digital image signal into a second digital image signal having a second resolution lower than said first resolution;

predicting means including class categorizing means for processing reference pixels of said second digital image signal to produce a class code, and means for producing a predicted signal at said first resolution from said reference pixels and said class code; and

means for subtracting said predicted signal from said first digital image signal to produce a differential signal.

2. The apparatus of claim 1, further comprising means for encoding said differential signal and said second digital image signal to produce respective compressed image signals.

3. The apparatus of claim 1, wherein said reference pixels are adjacent in at least one of a spatial dimension and a temporal dimension.

4. The apparatus of claim 1, wherein said class categorizing means is operative to compress said reference pixels to produce said class code.

5. The apparatus of claim 4, wherein said categorizing means is operative using one of adaptive dynamic range coding, differential pulse code modulation and block truncation coding.

6. The apparatus of claim 1, wherein said means for producing a predicted signal includes memory means for reading out at least one value corresponding to said class code, and combining means for combining the at least one read out value and said reference pixels to generate said predicted signal.

7. The apparatus of claim 6, wherein said memory means is operative to store predictive coefficients, and said combining means is operative to respectively multiply the predictive coefficients read out from said memory means by said reference pixels and to sum the results of the multiplications to produce a pixel of said predicted signal.

8. The apparatus of claim 6, wherein said memory means is operative to store representative values, and said combining means is operative to obtain a dynamic range and redundant component of said reference pixels, and to combine said dynamic range, said redundant component and the representative value read out from said memory means to produce a pixel of said predicted signal.

9. An apparatus for decoding first and second digital image signals having first and second resolutions, said first resolution being higher than said second resolution, comprising:

predicting means including class categorizing means for processing reference pixels of said second digital image signal to produce a class code, and means for producing a predicted signal at said first resolution from said reference pixels and said class code; and

means for adding said first digital image signal and said predicted signal to produce an output signal.

10. The apparatus of claim 9, further comprising means for decoding first and second encoded digital image signals to respectively produce said first and second digital image signals.

11. The apparatus of claim 9, wherein said reference pixels are adjacent in at least one of a spatial dimension and a temporal dimension.

12. The apparatus of claim 9, wherein said class categorizing means is operative to compress said reference pixels to produce said class code.

13. The apparatus of claim 12, wherein said categorizing means is operative using one of adaptive dynamic range coding, differential pulse code modulation and block truncation coding.

14. The apparatus of claim 9, wherein said means for producing a predicted signal includes memory means for reading out at least one value corresponding to said class code, and combining means for combining the at least one read out value and said reference pixels to generate said predicted signal.

15. The apparatus of claim 14, wherein said memory means is operative to store predictive coefficients, and said combining means is operative to respectively multiply the predictive coefficients read out from said memory means by said reference pixels and to sum the results of the multiplications to produce a pixel of said predicted signal.

16. The apparatus of claim 14, wherein said memory means is operative to store representative values, and said combining means is operative to obtain a dynamic range and redundant component of said reference pixels, and to combine said dynamic range, said redundant component and the representative value read out from said memory means to produce a pixel of said predicted signal.

17. A method of encoding a first digital image signal representing pixels and having a first resolution, comprising the steps of:

forming said first digital image signal into a second digital image signal having a second resolution lower than said first resolution;

processing reference pixels of said second digital image signal to produce a class code;

producing a predicted signal at said first resolution from said reference pixels and said class code; and

subtracting said predicted signal from said first digital image signal to produce a differential signal.

18. A method for decoding first and second digital image signals having first and second resolutions, said first resolution being higher than said second resolution, comprising the steps of:

processing reference pixels of said second digital image signal to produce a class code;

producing a predicted signal at said first resolution from said reference pixels and said class code; and

adding said first digital image signal and said predicted signal to produce an output signal.

19. A method of generating predictive coefficients for a digital image signal, comprising the steps of:

forming learning data representing an image into blocks;

determining a class code for each of said blocks;

defining a selected pixel in each of said blocks having the same class code as a prediction of other pixels in said each of said blocks respectively multiplied by said predictive coefficients;

generating values for said predictive coefficients which minimize the difference between the actual values of said selected pixel and the predicted values of said selected pixel in each of said blocks having the same class code; and

associating the values for said predictive coefficients with the respective class code.

20. A method according to claim 19, further comprising the step of eliminating blocks having a dynamic range lower than a predetermined threshold.

21. A method according to claim 19, wherein said block comprises pixels adjacent in at least one of a spatial dimension and a temporal dimension.

22. A method of generating representative values for a digital image signal, comprising the steps of:

forming learning data representing an image into blocks;

determining a class code for each of said blocks;

eliminating a redundant component from a selected pixel of each of said blocks;

normalizing the redundancy eliminated selected pixel of each of said blocks by a dynamic range of the respective blocks to produce respective sample values;

averaging said sample values for those of said blocks having the same class code to produce a representative value; and

associating said representative value with the respective class code.

23. A method according to claim 22, further comprising the step of eliminating blocks having a dynamic range lower than a predetermined threshold.

24. A method according to claim 22, wherein said block comprises pixels adjacent in at least one of a spatial dimension and a temporal dimension.