Multiscale coding of images

• US 4,987,480 A
• Filed: 07/11/1989
• Issued: 01/22/1991
• Est. Priority Date: 07/11/1989
• Status: Expired due to Term

• Current Assignee: Massachusetts Institute of Technology
• Original Assignee: Massachusetts Institute of Technology
• Inventors: Butera, William J., Lippman, Andrew B.
• Primary Examiner(s): Britton, Howard W.
• Application Number:
  US07/378,071
• Time in Patent Office: 560 Days
• Field of Search: 358/133, 358/13, 358/12
• US Class Current: 348/396.1
• CPC Class Codes: H04N 11/042 Codec means
• CPC Class Codes: H04N 19/186 the unit being a colour or a chrominance component
• CPC Class Codes: H04N 19/61 in combination with predictive coding
• CPC Class Codes: H04N 19/619 the transform being operated outside the prediction loop
• CPC Class Codes: H04N 19/62 by frequency transforming in three dimensions H04N19/63 takes precedence
• CPC Class Codes: H04N 19/63 using sub-band based transform, e.g. wavelets
• CPC Class Codes: H04N 19/94 Vector quantisation

Abstract

Image encoding apparatus and methods include recursively decomposing the luminance values of sequential frames into spatio-temporal spectral components, by quadrature mirror filter (QMF) processing of the luminance values. The filtered signals are subsampled, and each decomposed level of the spectral components is coded by vector quantization.

Claims

1. A method of encoding an image, the image including a plurality of frames defined by a set of chrominance and luminance values, the method comprising the steps of

recursively decomposing, to at least two levels of recursion, the luminance values of sequential frames into spatio-temporal spectral components, said spatio-temporal spectral components including a plurality of spatial components and a plurality of temporal components,

said decomposing step including the steps of

quadrature mirror filtering signals representative of said luminance values and

subsampling said filtered signals, and

coding each recursively decomposed level of at least one of said spectral components by vector quantization.

2. The method of claim 1 wherein the step of decomposing the luminance into spectral components further comprises separating the luminance values into high and low horizontal frequency subbands.

3. The method of claim 1 wherein the step of decomposing the luminance into spectral components further comprises separating the luminance values into high and low vertical frequency subbands.

4. The method of claim 1 wherein the step of decomposing the luminance into spectral components further comprises separating the luminance values into high and low temporal frequency subbands.

5. The method of claim 1 comprising the further step of discarding at least one spectral component corresponding to diagonal image features.

6. The method of claim 1 wherein selected spectral components are transmitted without vector quantization to reduce bandwidth requirements.

7. The method of claim 1 wherein said decomposing step is executed on frame sets having at least two frames.

8. The method of claim 1 comprising the further steps of

decomposing the chrominance values of sequential frames into at least two chrominance spectral components, and

coding said at least two chrominance spectral components by vector quantization.

9. The method of claim 1 wherein the step of decomposing the luminance into spectral components further comprises separating the luminance values into high and low subbands for horizontal, vertical and temporal information.

10. The method of claim 9 wherein the method further comprises decomposing the subband containing low frequency horizontal, vertical and temporal information into finer high and low subbands of horizontal and vertical information.

11. The method of claim 5 comprising the further step of replacing the subband containing low frequency horizontal, low frequency vertical, and high frequency temporal information with a prediction error signal, said replacing step including the step of computing said prediction error signal by motion compensated interpolation between successive values of the subband corresponding to low horizontal, vertical and temporal information.

12. The method of claim 1 wherein said coding step includes the step of recursively constructing code tables for a given level from code tables utilized for preceding levels.

13. The method of claim 12 wherein said step of recursively constructing code tables includes the step of utilizing a different block size at each level.

14. The method of claim 12 wherein the step of constructing code tables includes the step of constructing orientation-independent code tables.

15. The method of claim 12 wherein the step of constructing code tables includes the step of controlling code table size by establishing address pointers to indirectly address code tables.

16. The method of claim 12 comprising the further step of entropy coding said code tables to reduce bandwidth requirements.

17. The method of claim 12 comprising the further step of transmitting code table update information during scene boundary periods.

18. The method of claim 17 wherein the step of transmitting code table update information includes the step of lookahead processing to determine any of scene boundaries or image sequence locations where bandwidth can be allocated to table update information.

19. Apparatus for encoding an image, the image including a plurality of frames defined by a set of chrominance and luminance values, the apparatus comprising

decomposing means for recursively decomposing, to at least two levels of recursion, the luminance values of sequential frames into spatio-temporal spectral components, said spatio-temporal spectral components including a plurality of spatial components and a plurality of temporal components,

said decomposing means including

quadrature mirror filter means for filtering signals representative of said luminance values and

subsampling means for subsampling said filtered signals, and

vector quantization means for coding each recursively decomposed level of at least one of said spectral components by vector quantization.

20. Apparatus according to claim 19 wherein the decomposing means includes means for separating the luminance values into high and low horizontal frequency subbands.