Segmented-block coding
The current application is directed to flexible methods for motion-vector-based encoding of macroblocks within video frames. The flexible methods for encoding video-frame macroblocks provide for segmentation of a video-frame block and encoding the segmented video-frame block by a segmented-block motion vector that includes a reference to a segmentation-defining region of a segmentation map and that also includes references to sources of intensity-and-color data for each segment. Segmented-block motion vectors provide for flexible segmentation-based encoding of video-frame blocks without sacrificing the coding efficiencies attendant with conventional motion-vector-based video encoding.
1. A video-processing device comprising:
a processor;
a memory device; and
computer instructions stored within the memory device that control the processor to:
determine a motion vector for a macroblock of a frame of video based on a function of the macroblock and reference macroblocks of the frame;
determine, based on information available to an encoder of the video and a decoder of the video, a segmentation map that partitions the macroblock into a plurality of segments along one or more q-pel resolution boundaries of the segments, wherein each of the segments represents an object in the frame of video;
determine a plurality of segmented-block motion vectors for the macroblock based on the segmentation map; and
encode the macroblock using the segmented-block motion vectors.
2. The video-processing device of claim 1 , wherein the function comprises a mean squared difference between the macroblock and the reference macroblocks.
3. The video-processing device of claim 1 , the segmentation map comprising macroblock regions having a consistent trajectory, the consistent trajectory being consistent with a trajectory of the macroblock.
4. The video-processing device of claim 3 , wherein the segmented-block motion vectors are determined by instructions that control the processor to:
identify a lowest depth region of the frame, the lowest depth region comprising neighboring pixels of a similar depth;
identify macroblocks of the macroblock regions that have possible trajectories consistent with the consistent trajectory;
identify a segmentation macroblock of the identified macroblocks; and
generate the segmented-block motion vectors to comprise a plurality of motion vectors for the segmentation macroblock.
5. The video-processing device of claim 4 , wherein the segmented-block motion vectors are determined by instructions that control the processor to generate the segmented-block motion vectors to comprise a motion vector for a suitable macroblock of each of the segments.
6. The video-processing device of claim 1 , wherein the instructions control the processor to:
determine a storage size of the macroblock encoded based on the segmented-block motion vectors; and
determine the storage size of the macroblock encoded based on the motion vector.
7. The video-processing device of claim 6 , wherein the instructions control the processor to encode the macroblock using the motion vector if the storage size of the macroblock encoded based on the segmented-block motion vectors is greater than the storage size of the macroblock encoded based on the motion vector.
8. A method for encoding a macroblock of a frame of video, the method comprising:
determining a motion vector for a macroblock of the frame of video based on a function of the macroblock and reference macroblocks of the frame, wherein the function comprises a mean squared difference between the macroblock and the reference macroblocks;
determining, based on information available to an encoder of the video and a decoder of the video, a segmentation map that partitions the macroblock into a plurality of segments along one or more q-pel resolution boundaries of the segments, wherein each of the segments represents an object in the frame of video;
determining a plurality of segmented-block motion vectors for the macroblock based on the segmentation map; and
encoding the macroblock using the segmented-block motion vectors.
9. The method of claim 8 , the segmentation map comprising macroblock regions having a consistent trajectory, the consistent trajectory being consistent with a trajectory of the macroblock.
10. The method of claim 9 , wherein the segmented-block motion vectors are determined by:
identifying a lowest depth region of the frame, the lowest depth region comprising neighboring pixels of a similar depth;
identifying macroblocks of the macroblock regions that have possible trajectories consistent with the consistent trajectory;
identifying a segmentation macroblock of the identified macroblocks; and
generating the segmented-block motion vectors to comprise a motion vector for the segmentation macroblock.
11. The method of claim 10 , wherein the segmented-block motion vectors are determined by generating the segmented-block motion vectors to comprise a motion vector for a suitable macroblock of each of the segments.
12. The method of claim 8 , comprising generating a segmentation map of the frame, the segmentation map comprising segmented macroblock regions.
13. The method of claim 8 , comprising:
determining a storage size of the macroblock encoded based on the segmented-block motion vectors; and
determining a storage size of the macroblock encoded based on the motion vector.
14. The method of claim 13 , comprising encoding the macroblock using the motion vector if the storage size of the macroblock encoded based on the segmented-block motion vector is greater than the storage size of the macroblock encoded based on the motion vector.
15. A computer-readable storage memory device for storing computer-readable instructions, the computer-readable instructions providing a macroblock encoding system when executed by one or more processing devices, the computer-readable instructions comprising code configured to:
determine a motion vector for a macroblock of a frame of video based on a function of the macroblock and reference macroblocks of the frame, wherein the function comprises a mean squared difference between the macroblock and the reference macroblocks;
determine, based on information available to an encoder of the video and a decoder of the video, a segmentation map that partitions the macroblock into a plurality of segments along one or more q-pel resolution boundaries of the segments, wherein each of the segments represents an object in the frame of video;
determine a plurality of segmented-block motion vectors for the macroblock based on the segmentation map; and
encode the macroblock using the segmented-block motion vectors.
16. The computer-readable storage memory device of claim 15 , comprising instructions to determine the segmented-block motion vectors by:
identifying a lowest depth region of the frame, the lowest depth region comprising neighboring pixels of a similar depth;
identifying macroblocks of the segmented macroblock regions that have possible trajectories consistent with the consistent trajectory;
identifying a segmentation macroblock of the identified macroblocks; and
generating the segmented-block motion vectors to comprise a motion vector for the segmentation macroblock.
17. The computer-readable storage memory device of claim 16 , wherein the segmented-block motion vectors are determined by generating the segmented-block motion vectors to comprise a motion vector for a suitable macroblock of each of the segments.
18. The computer-readable storage memory device of claim 17 , comprising instructions to:
determine a storage size of the macroblock encoded based on the segmented-block motion vector; and
determine a storage size of the macroblock encoded based on the motion vector.
19. The computer-readable storage memory device of claim 18 , comprising instructions to encode the macroblock using the motion vector if the storage size of the macroblock encoded based on the segmented-block motion vector is greater than the storage size of the macroblock encoded based on the motion vector.
20. The computer-readable storage memory device of claim 18 , comprising instructions to include an indicator in the segmented-block motion vector to indicate that intrablock encoding is used for a specific segment.