Method and system of performing inter-frame prediction in video compression
Systems, devices, and methods of performing inter-frame prediction in video compression. A method includes: partitioning a video frame into superblocks, each superblock includes a plurality of macroblocks; determining an upper boundary on attributes that are associated with residual blocks of the plurality of macroblocks, based on pre-defined constraints on the total number of bits allocated for motion vector descriptions in a superblock; for each macroblock in a superblock, determining a partition size and a motion vector resolution which yield, for all macroblocks in the superblock, a motion vector description and a corresponding residual with attributes less than or equal to the upper boundary.
1. A method of performing inter-frame prediction in video compression, the method comprising:
partitioning a video frame into superblocks, each superblock comprising a plurality of macroblocks;
determining an upper boundary on attributes associated with residual blocks of said plurality of macroblocks based on pre-defined constraints on a total number of bits that are allocated for descriptors of motion vectors in a superblock; and
for each macroblock in a superblock, determining a partition size and a motion vector resolution which yield, for all macroblocks in said superblock, descriptors of motion vectors and a corresponding residual block with attributes that are less than or equal to said upper boundary.
2. The method of claim 1 ,
wherein each macroblock of the plurality of macroblocks comprises 16×16 pixels.
3. The method of claim 1 ,
wherein the attributes comprise a residual complexity.
4. The method of claim 1 ,
wherein the attributes comprise a measure of a total energy in the residual blocks.
5. The method of claim 1 ,
wherein the attributes comprise a measure of entropy of quantized Discrete Cosine Transform (DCT) taps of the residual blocks.
6. The method of claim 1 ,
wherein said partition size comprises any one of:
16×16 pixel size,
16×8 pixel size,
8×16 pixel size,
8×8 pixel size,
8×4 pixel size,
4×8 pixel size,
4×4 pixel size.
7. The method of claim 1 , comprising:
determining a number of bits required to code a motion vector for each of the block partitionings based on predetermined pixel accuracies.
8. The method of claim 1 ,
wherein said motion vector resolution comprises any one of:
a full pixel,
a half pixel,
a quarter pixel.
9. The method of claim 1 , comprising:
determining a residual complexity in a residual block associated with each of the block partitionings.
10. The method of claim 1 , comprising:
for each block partitioning, comparing a number of bits required to code a motion vector and a residual complexity value associated with the residual block.
11. The method of claim 10 , comprising:
discarding all block partitionings that are associated with motion vectors requiring the same number of bits for coding, except for a block partitioning with a least residual complexity value.
12. The method of claim 1 , comprising:
assigning an upper boundary requirement to residual complexity values of residual blocks.
13. An encoder for generating fixed-rate motion vectors for video compression, configured to perform the following steps:
partition a video frame into superblocks, each superblock comprising a plurality of macroblocks;
determine an upper boundary on attributes associated with residual blocks of said plurality of macroblocks based on pre-defined bit constraints on a total number of bits that are allocated for descriptors of motion vectors in a superblock; and
for each macroblock in a superblock, determine a partition size and a motion vector resolution which yield, for all macroblocks in said superblock, descriptors of motion vectors and a corresponding residual block with attributes that are less than or equal to said upper boundary.
14. The encoder of claim 13 ,
wherein each macroblock of the plurality of macroblocks comprises 16×16 pixels.
15. The encoder of claim 13 ,
wherein the attributes comprise a residual complexity.
16. The encoder of claim 13 ,
wherein the attributes comprise a measure of a total energy in the residual blocks.
17. The encoder of claim 13 ,
wherein the attributes comprise a measure of entropy of quantized Discrete Cosine Transform (DCT) taps of the residual blocks.
18. The encoder of claim 13 ,
wherein said partition size comprises any one of:
16×16 pixel size,
16×8 pixel size,
8×16 pixel size,
8×8 pixel size,
8×4 pixel size,
4×8 pixel size,
4×4 pixel size.
19. The encoder of claim 13 ,
wherein the encoder is further configured to determine a number of bits required to code a motion vector for each of the block partitionings based on a predetermined motion vector resolution.
20. The encoder of claim 13 ,
wherein said motion vector resolution comprises any one of:
a full pixel,
a half pixel,
a quarter pixel.
21. The encoder of claim 13 ,
wherein the encoder is further configured to determine a residual complexity in a residual block associated with each of said block partitionings.
22. The encoder of claim 21 ,
wherein the encoder is further configured to compare, for each block partitioning, a number of bits required to code a motion vector and a residual complexity value associated with the residual block.
23. The encoder of claim 13 ,
wherein the encoder is further configured to discard all block partitionings that are associated with motion vectors requiring the same number of bits for coding except for a block partitioning with a least residual complexity value.
24. The encoder of claim 13 ,
wherein the encoder is further configured to assign an upper boundary requirement to residual complexity values of residual blocks.