Three-dimensional wavelet-based scalable video compression

• US 20040028138A1
• Filed: 04/23/2003
• Published: 02/12/2004
• Est. Priority Date: 10/24/2000
• Status: Active Grant

• Current Assignee: Anyconnect Private Limited
• Original Assignee: Eyeball Networks Incorporated
• Inventors: Vass, Jozsef, Khan, Shahadatullah, Piche, Christopher
• Granted Patent: US 6,931,068 B2
• Time in Patent Office: Days
• US Class Current: 375/240.19
• CPC Class Codes: H04N 19/13 Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
• CPC Class Codes: H04N 19/61 in combination with predictive coding
• 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 19/647 using significance based coding, e.g. Embedded Zerotrees of Wavelets [EZW] or Set Partitioning in Hierarchical Trees [SPIHT]

Abstract

A method of encoding an input video signal for communication over a computer network, the method comprising the steps of: i) applying a three-dimensional wavelet-based temporal and spatial decomposition to produce a plurality of spatio-temporal subbands; ii) quantizing the coefficients of the spatio-temporal subbands with a uniform scalar quantizer to produce a significance map; and iii) ran-length and adaptive arithmetic coding of the signal by encoding the significance map, encoding the signs of all significant coefficients, and encoding the magnitudes of significant coefficients, in bit-plane order starting with the most significant bit-plane.

Claims

1. A method of encoding an input video signal comprising a group of video frames for communication over a computer network, the method comprising the steps of:

i) applying a two-level temporal decomposition using a wavelet to said group of frames to produce a plurality of temporal subbands;

ii) applying a spatial decomposition to each said temporal subband to produce a plurality of spatio-temporal subbands;

iii) quantizing the coefficients of said spatio-temporal subbands with a uniform scalar quantizer to produce a significance map; and

iv) run-length and adaptive arithmetic coding of said signal by

a) encoding the significance map through a combination of run-length and adaptive arithmetic coding by;

A) coding run-length codewords using N-ary adaptive arithmetic coding, where N is the maximum run-length previously observed in coding the significance map; and

B) encoding the current run-length codeword using one of a specified plurality of probability models, the probability model being selected by a rule which selects a probability model according to the previous run-length codeword;

b) encoding the signs of all significant coefficients using the number of significant coefficients in a four pixel neighborhood as context; and

c) encoding the magnitudes of significant coefficients, in bit-plane order starting with the most significant bit-plane, using the number of significant coefficients in a four pixel neighborhood as context.

2. The method of claim 1 wherein, in step iv) a) B., the number of said probability models is five.

3. The method of claim 1 wherein said group of frames comprises four frames.

4. The method of claim 1 wherein the Haar wavelet is used for temporal decomposition.

5. The method of claim 1 wherein the Daubechies 9/7 filter is used for spatial decomposition.

6. A computer program product for encoding an input video signal comprising a group of video frames for communication over a computer network, said computer program product comprising:

i) a computer usable medium having computer readable program code means embodied in said medium for;

a) applying a two-level temporal decomposition using a wavelet to said group of frames to produce a plurality of temporal subbands;

b) applying a spatial decomposition to each said temporal subband to produce a plurality of spatio-temporal subbands;

c) quantizing the coefficients of said spatio-temporal subbands with a uniform scalar quantizer to produce a significance map; and

d) run-length and adaptive arithmetic coding of said signal by

A) encoding the significance map through a combination of run-length and adaptive arithmetic coding by;

i) coding run-length codewords using N-ary adaptive arithmetic coding, where N is the maximum run-length previously observed in coding the significance map; and

ii) encoding the current run-length codeword using one of a specified plurality of probability models, the probability model being selected by a rule which selects a probability model according to the previous run-length codeword;

B) ending the signs of all significant coefficients using the number of significant coefficients in a four pixel neighborhood as context; and

C) encoding the magnitudes of significant coefficients, in bit-plane order starting with the most significant bit-plane, using the number of significant coefficients in a four pixel neighborhood as context.

7. The computer program product of claim 6 wherein the number of said probability models is five.

8. The computer program product of claim 6 wherein said group of frames comprises four frames.

9. The computer program product of claim 6 wherein the Haar wavelet is used for temporal decomposition.

10. The computer program product of claim 6 wherein the Daubechies 9/7 filter is used for spatial decomposition.

11. An article comprising:

i) a computer readable modulated carrier signal;

ii) means embedded in said signal for encoding an input video signal comprising a group of video frames for communication over a computer network, said means comprising means for;

a) applying a two-level temporal decomposition using a wavelet to said group of frames to produce a plurality of temporal subbands;

b) applying a spatial decomposition to each said temporal subband to produce a plurality of spatio-temporal subbands;

c) quantizing the coefficients of said spatio-temporal subbands with a uniform scalar quantizer to produce a significance map; and

d) run-length and adaptive arithmetic coding of said signal by;

A) encoding the significance map;

i) coding run-length codewords using N-ary adaptive arithmetic coding, where N is the maximum run-length previously observed in coding the significance map; and

ii) encoding the current run-length codeword using one of a specified plurality of probability models, the probability model being selected by a rule which selects a probability model according to the previous run-length codeword;

B) ending the signs of all significant coefficients using the number of significant coefficients in a four pixel neighborhood as context; and

C) encoding the magnitudes of significant coefficients, in bit-plane order starting with the most significant bit-plane, using the number of significant coefficients in a four pixel neighborhood as context.

12. The article of claim 11 wherein the number of said probability models is five.

13. The article of claim 11 wherein said group of frames comprises four frames.

14. The article of claim 11 wherein the Haar wavelet is used for temporal decomposition.

15. The article of claim 11 wherein the Daubechies 9/7 filter is used for spatial decomposition.