Hierarchical Syntax Packing for Video Motion Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently encoding video frames on parallel processing systems, particularly in reducing the computational complexity of motion estimation while maintaining video quality.
Innovation Solution
The proposed solution involves a hierarchical packing approach for syntax elements, which includes identifying candidate search regions in a reference frame, performing refinement searches within these regions, and using a voting scheme to select the best motion vectors for further refinement. This method optimizes the search for the best match by limiting the number of candidates and using parallel processing to enhance efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion estimation is performed by searching all possible regions in the reference frame, then the accuracy of motion vectors is improved, but the computational complexity increases significantly
Solution Approach 1:
The patent divides the reference frame into multiple candidate search regions based on motion vectors from previously decoded macroblocks. Instead of searching the entire reference frame, the algorithm segments the search space into localized regions where motion is more likely to occur, thereby reducing computational complexity while maintaining motion estimation accuracy.
Solution Approach 2:
The patent applies different search strategies to different regions of the reference frame. Candidate search regions are identified based on local motion characteristics from neighboring macroblocks, allowing the algorithm to concentrate computational resources in areas with significant motion while using simpler methods in static regions.
2Productivity
If the search region is reduced to decrease computational complexity, then the processing speed is improved, but the motion estimation accuracy may deteriorate
Solution Approach 1:
The patent performs preliminary identification of candidate search regions using motion vectors from already decoded neighboring macroblocks before conducting the full motion estimation search. This preliminary action narrows down the search space to regions most likely to contain the best match, ensuring that the reduced search region does not compromise motion estimation accuracy.
Solution Approach 2:
The algorithm uses feedback from previously decoded macroblocks to dynamically adjust and identify candidate search regions. Motion vectors from neighboring blocks provide feedback about the likely location of matching regions, allowing the search to be concentrated in appropriate areas while maintaining accuracy.
3Productivity
If parallel processing is used to increase processing speed, then the productivity is improved, but the device complexity increases
Solution Approach 1:
The patent segments the motion estimation process into independent tasks that can be executed in parallel. Different macroblocks or different candidate search regions can be processed simultaneously by multiple processing elements, increasing productivity while keeping each individual processing task relatively simple.
Data Source
AI summary
A system for improving video coding performance while using a merge mode in motion estimation. The system comprises a processor configured to perform one or more refinement searches on a plurality of candidate regions of a current frame, wherein the plurality of candidate regions comprises a candidate region identified in a reference frame and a plurality of candidate search regions, and wherein the one or more refinement searches reduce the plurality of candidate regions to obtain reduced candidates.