System and method for enhanced DMVD processing

• US 9,509,995 B2
• Filed: 12/21/2010
• Issued: 11/29/2016
• Est. Priority Date: 12/21/2010
• Status: Active Grant

• Current Assignee: Tahoe Research Limited (f/k/a Learndale Limited) (Vector Capital Corporation)
• Original Assignee: Intel Corporation
• Inventors: Xu, Lidong, Chiu, Yi-Jen, Zhang, Wenhao, Jiang, Hong
• Primary Examiner(s): Bailey, Frederick
• Assistant Examiner(s): Nawaz, Talha M
• Application Number:
  US13/996,582
• Publication Number:
  US 20140146890A1
• Time in Patent Office: 2,170 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: H04N 19/436 using parallelised computational arrangements
• CPC Class Codes: H04N 19/44 Decoders specially adapted therefor, e.g. video decoders which are asymmetric with respect to the encoder
• CPC Class Codes: H04N 19/513 Processing of motion vectors

Abstract

To let decoder side motion vector derivation (DMVD) coded blocks be decoded in parallel, decoder side motion estimation (ME) dependency on spatially neighboring reconstructed pixels can be removed. Mirror ME and projective ME are only performed on two reference pictures, and the spatially neighboring reconstructed pixels will not be considered in the measurement metric of the decoder side ME. Also, at a video decoder, motion estimation for a target block in a current picture can be performed by calculating a motion vector for a spatially neighboring DMVD block, using the calculated motion vector to predict motion vectors of neighboring blocks of the DMVD block, and decoding the DMVD block and the target block in parallel. In addition, determining a best motion vector for a target block in a current picture can be performed by searching only candidate motion vectors in a search window, wherein candidate motion vectors are derived from a small range motion search around motion vectors of neighboring blocks.

Claims

1. A method, comprising:

receiving a decoder-side motion vector derivation (DMVD) encoded block and a non-DMVD encoded block;

determining the DMVD encoded block is spatially neighboring the non-DMVD encoded block;

for the DMVD encoded block, performing motion estimation (ME) using temporally neighboring reconstructed pixels in reference pictures, without using spatially neighboring reconstructed pixels in a current picture;

for the non-DMVD encoded block, predicting a motion vector of the non-DMVD encoded block based, at least in part, on motion vectors of spatially neighboring blocks, and recovering a motion vector difference from received data, wherein said predicting comprises at least one of;

limiting the spatially neighboring blocks to an available subset of the spatially neighboring blocks that excludes the DMVD block as unavailable for motion vector prediction for the non-DMVD encoded block, or

calculating a motion vector of the DMVD encoded block based on an available subset of blocks spatially neighboring the DMVD block and using the calculated motion vector in the motion vector predicting for the non-DMVD encoded block before completing decoder-side motion estimation of the DMVD encoded block; and

decoding the DMVD encoded block and the non-DMVD encoded block in parallel.

2. The method of claim 1, wherein said calculating of a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

calculating a weighted average of motion vectors of the available blocks that spatially neighbor the neighboring DMVD block.

3. The method of claim 1, wherein said calculating of a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

calculating a median filtered value for motion vectors of the available blocks that spatially neighbor the neighboring DMVD block.

4. The method of claim 1, wherein said calculating of a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

calculating a weighted average of scaled motion vectors of the available blocks that temporally neighbor the neighboring DMVD encoded block.

5. The method of claim 1, wherein said calculating of a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

calculating a median filtered value of scaled motion vectors of the available blocks that temporally neighbor the neighboring DMVD block.

6. The method of claim 1, wherein said calculating of a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

calculating a weighted average of

motion vectors of the available blocks that spatially neighbor the neighboring DMVD block, and

motion vectors of the available blocks that temporally neighbor the neighboring DMVD block.

7. The method of claim 1, wherein said calculating of a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

calculating a median filtered value of

motion vectors of the available blocks that spatially neighbor the neighboring DMVD block, and

motion vectors of the available blocks that temporally neighbor the neighboring DMVD block.

8. The method of claim 1, wherein said calculating of a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

performing a fast candidate-based motion estimation, by searching in a small range for a best candidate motion vector for the DMVD block, and choosing the best candidate motion vector on the basis of whether the candidate motion vector generates a minimal sum of absolute differences.

9. A system, comprising:

a processor; and

a memory in communication with said processor, for storing a plurality of processing instructions for directing said processor to

receive a decoder-side motion vector derivation (DMVD) encoded block and a non-DMVD encoded block;

determine the DMVD encoded block is spatially neighboring the non-DMVD encoded block;

for the DMVD encoded block, perform motion estimation (ME) using temporally neighboring reconstructed pixels in reference pictures, without using spatially neighboring reconstructed pixels in a current picture;

for the non-DMVD encoded block, predict a motion vector of the non-DMVD encoded block based, at least in part, on motion vectors of spatially neighboring blocks, and recover a motion vector difference from received data, wherein said predict comprises at least one of;

a limit on the spatially neighboring blocks to an available subset of the spatially neighboring blocks that excludes the DMVD block as unavailable for motion vector prediction for the non-DMVD encoded block, or

a calculation of a motion vector of the DMVD encoded block based on an available subset of blocks spatially neighboring the DMVD block and a use of the calculated motion vector in the motion vector prediction for the non-DMVD encoded block before decoder-side motion estimation of the DMVD encoded block is completed; and

decode the DMVD encoded block and the non-DMVD encoded block in parallel.

10. The system of claim 9, wherein said processing instructions for directing said processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprise:

processing instructions for directing said processor to calculate a weighted average of motion vectors of the available blocks that spatially neighbor the neighboring DMVD block.

11. The system of claim 9, wherein said processing instructions for directing said processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprise:

processing instructions for directing said processor to calculate a median filtered value for motion vectors of the available blocks that spatially neighbor the neighboring DMVD block.

12. The system of claim 9, wherein said processing instructions for directing said processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprise:

processing instructions for directing said processor to calculate a weighted average of scaled motion vectors of the available blocks that temporally neighbor the neighboring DMVD encoded block.

13. The system of claim 9, wherein said processing instructions for directing said processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprise:

processing instructions for directing said processor to calculate a median filtered value of scaled motion vectors of the available blocks that temporally neighbor the neighboring DMVD block.

14. The system of claim 9, wherein said processing instructions for directing said processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprise:

processing instructions for directing said processor to calculate a weighted average of

motion vectors of the available blocks that spatially neighbor the neighboring DMVD block, and

motion vectors of the available blocks that temporally neighbor the neighboring DMVD block.

15. The system of claim 9, wherein said processing instructions for directing said processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprise:

processing instructions for directing said processor to calculate a median filtered value of

motion vectors of the available blocks that spatially neighbor the neighboring DMVD block, and

motion vectors of the available blocks that temporally neighbor the neighboring DMVD block.

16. The system of claim 9, wherein said processing instructions for directing said processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprise:

processing instructions for directing said processor to perform a fast candidate-based motion estimation, by searching in a small range for a best candidate motion vector for the DMVD block, and choosing the best candidate motion vector on the basis of whether the candidate motion vector generates a minimal sum of absolute differences.

17. A computer program product including a non-transitory computer readable medium having computer program logic stored therein, the computer program logic comprising:

logic to cause a processor to receive a decoder-side motion vector derivation (DMVD) encoded block and a non-DMVD encoded block;

logic to cause the processor to determine the DMVD encoded block is spatially neighboring the non-DMVD encoded block;

logic to cause the processor, for the DMVD encoded block, to perform motion estimation (ME) using temporally neighboring reconstructed pixels in reference pictures, without using spatially neighboring reconstructed pixels in a current picture;

logic to cause the processor, for the non-DMVD encoded block, to predict a motion vector of the non-DMVD encoded block based, at least in part, on motion vectors of spatially neighboring blocks, and to recover a motion vector difference from received data wherein said predict comprises at least one of;

a limit on the spatially neighboring blocks to an available subset of the spatially neighboring blocks that excludes the DMVD block as unavailable for motion vector prediction for the non-DMVD encoded block, or

a calculation of a motion vector of the DMVD encoded block based on an available subset of blocks spatially neighboring the DMVD block and a use of the calculated motion vector in the motion vector prediction for the non-DMVD encoded block before decoder-side motion estimation of the DMVD encoded block is completed; and

logic to cause the processor to decode the DMVD encoded block and the non-DMVD encoded block in parallel.

18. The computer program product of claim 17, wherein the computer program logic to cause the processor to calculate of a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

logic to cause the processor to calculate a weighted average of motion vectors of the available blocks that spatially neighbor the neighboring DMVD block.

19. The computer program product of claim 17, wherein the computer program logic to cause the processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

logic to cause the processor to calculate a median filtered value for motion vectors of the available blocks that spatially neighbor the neighboring DMVD block.

20. The computer program product of claim 17, wherein the computer program logic to cause the processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

logic to cause the processor to calculate a weighted average of scaled motion vectors of the available blocks that temporally neighbor the neighboring DMVD encoded block.

21. The computer program product of claim 17, wherein the computer program logic to cause the processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

logic to cause the processor to calculate a median filtered value of scaled motion vectors of blocks that temporally neighbor the neighboring DMVD block.

22. The computer program product of claim 17, wherein the computer program logic to cause the processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

logic to cause the processor to calculate a weighted average of

motion vectors of the available blocks that spatially neighbor the neighboring DMVD block, and

motion vectors of the available blocks that temporally neighbor the neighboring DMVD block.

23. The computer program product of claim 17, wherein the computer program logic to cause the processor to calculate a motion vector of the DMVD encoded block that neighbors the non-DMVD encoded block comprises:

logic to cause the processor to calculate a median filtered value of

motion vectors of the available blocks that spatially neighbor the neighboring DMVD block, and

motion vectors of the available blocks that temporally neighbor the neighboring DMVD block.

24. The computer program product of claim 17, wherein the computer program logic to cause the processor to calculate a motion vector of a DMVD encoded block that neighbors the non-DMVD encoded block comprises:

logic to cause the processor to perform a fast candidate-based motion estimation, by searching in a small range for a best candidate motion vector for the DMVD block, and choosing the best candidate motion vector on the basis of whether the candidate motion vector generates a minimal sum of absolute differences.