Granted Patent B1

US 8,687,699 · App. 11/434,953 · Granted Apr 1, 2014

Abstract

Claims (33)

1. A method of motion estimation for video processing, said method comprising steps of:

(A) receiving a current block of a video signal comprising a series of frames, wherein each frame comprises a plurality of blocks;

(B) determining an optimal residual block having an optimal rate-distortion cost for the current block, further comprising steps of:

(1) generating a plurality of residual blocks by performing motion compensation on the current block;

(2) generating a plurality of rate-distortion costs for a plurality of generated bitstream segments using a first transform for each of the plurality of residual blocks;

(3) selecting the optimal residual block having the optimal rate-distortion cost from the plurality of residual blocks; and

(C) transforming said optimal residual block for the purpose of encoding using a second transform, wherein the second transform has a higher complexity than the first transform, said second transform being performed on the optimal residual block in a condition corresponding to the residual block before the first transform is performed on the optimal residual block, and said first and second transforms are associated with two distinct modes in a transform circuit.

2. The method according to claim 1 , wherein said method implements a comprehensive performance evaluation of one or more encoding decisions and/or parameters.

3. The method according to claim 1 , wherein the first transform for generating a plurality of rate-distortion costs comprises a Hadamard transform and the second transform directed to transforming said selected residual block with said optimal cost comprises a Discrete Cosine Transform (DCT).

4. The method according to claim 3 , wherein step (C) generates an H.264 compliant coded video signal.

5. The method according to claim 1 , further comprising the step of:

calculating a plurality of estimated distortions for said residual blocks, wherein (i) each one of said distortions is between one of reconstructed blocks after transforming said residual blocks, quantizing resulted transform coefficients, dequantization and inverse transform dequantized coefficients and a respective reference block from a reference frame and (ii) said rate-distortion costs are based on said estimated distortions.

6. The method according to claim 1 , further comprising the step of:

calculating a plurality of estimated bit rates for coded representation of said residual block by performing quantization and entropy coding of said low complexity transform that would have been performed on DCT coefficients.

7. The method according to claim 6 , further comprising the step of:

estimating the bit rate for the coded representation of the said residual block by applying prior art methods for estimating bit rate for representing quantized DCT coefficients to quantized low complexity transform coefficients.

8. The method according to claim 6 , further comprising the step of:

estimating the bit rate for the coded representation of the said residual block when context based adaptive entropy coding is applied.

9. The method according to claim 1 , wherein sub-step (2) in step (B) comprises the sub-step of:

calculating only a portion of each of a number of coefficient sets, wherein each of said portions comprises a respective at least one lowest frequency coefficient.

10. The method according to claim 1 , wherein sub-step (2) in step (B) comprises calculating a subset of coefficients of the said low complexity transform.

11. The method according to claim 10 , wherein the subset of coefficients are the four coefficients at the upper left corner of the coefficients in the transform domain.

12. The method according to claim 10 , wherein the subset of coefficients are estimated from the pixel domain using pixel domain characteristics without performing exact transform.

13. The method according to claim 10 , wherein the subset of coefficients are estimated with a linear combination of row sums of the residual signal.

14. The method according to claim 1 , wherein said method implements a comprehensive performance evaluation of one or more encoding decisions and/or parameters.

15. An apparatus comprising:

a motion estimation circuit configured to generate a plurality of residual blocks by performing motion compensation on a current block received from a video signal;

a transform circuit configured to generate a plurality of rate-distortion costs for a plurality of generated bitstream segments using a first transform for each of the plurality of residual blocks;

a rate distortion circuit configured to select an optimal block having the optimal rate-distortion cost from the plurality of residual blocks;

wherein the transform circuit is further configured to transform said optimal residual block for the purpose of encoding, using a second transform, wherein the second transform has a higher complexity than the first transform, said transform circuit configured to transform the optimal residual block using the second transform in a condition corresponding to the residual block before the first transform is performed on the optimal residual block, and said first and second transforms are associated with two distinct modes in a transform circuit.

16. The apparatus according to claim 15 , wherein said bitstream is H.264 compliant.

17. The apparatus according to claim 15 , wherein said first transform comprises a Hadamard transform and the second transform comprises a Discrete Cosine Transform (DCT).

18. The apparatus according to claim 15 , wherein said first transform has a lower complexity than a DCT transform.

Assignments (12)