Video Image Processing With Reduced ATMVP Candidate Scanning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing Advanced Temporal Motion Vector Prediction (ATMVP) technology in video coding standards suffers from redundant operations and sub-CU size settings that do not match the storage granularity, leading to inefficiencies and increased complexity.
Innovation Solution
A method to construct a motion vector candidate list by scanning a reduced number of candidate motion vectors, using a preset condition to determine a reference motion vector, and adaptively setting sub-CU size to 8×8, thereby reducing redundant operations and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the current ATMVP technology scans all candidate motion vectors to determine a time-domain vector, then the prediction accuracy is improved, but the computational complexity and processing time increase
Solution Approach 1:
The patent segments the candidate motion vector list into multiple parts and processes them in parallel. Instead of sequentially scanning all candidates, the method divides the search space into segments that can be evaluated simultaneously, reducing the overall processing time while maintaining comprehensive coverage of candidate vectors.
Solution Approach 2:
The patent performs preliminary filtering of candidate motion vectors based on preset conditions before the full scanning process. By pre-identifying and eliminating unlikely candidates based on motion characteristics and reference frame analysis, the method reduces the number of vectors that require detailed evaluation, thereby lowering computational complexity while preserving prediction accuracy.
2Measurement precision
If the sub-CU size is adaptively set at frame level with default 4×4 and increased to 8×8 under certain conditions, then the prediction precision is improved, but the storage granularity mismatch causes inefficiency
Solution Approach 1:
The patent changes the sub-CU size parameter from adaptive frame-level setting to a fixed configuration that aligns with storage granularity. By setting sub-CU size to 8×8 as a standard fixed value, the method eliminates the overhead of adaptive decision-making and storage granularity mismatch, improving encoding efficiency while maintaining sufficient prediction precision through other mechanisms.
3Measurement precision
If TMVP and ATMVP technologies are both applied, then the motion vector prediction accuracy is improved, but redundant operations occur in certain scenarios
Solution Approach 1:
The patent merges the TMVP and ATMVP technologies into a unified prediction framework. By integrating the temporal motion vector prediction mechanisms of both methods into a single coherent process, the patent eliminates redundant operations that occur when both technologies are applied separately, while maintaining the prediction accuracy benefits of combining temporal and spatial information.
Data Source
Figure 1~2
Figure 3~4
Figure 5~7
AI summary
A video image processing method is provided which includes scanning N neighboring blocks among M preset neighboring blocks of a current image block sequentially, determining a target neighboring block according to a scan result, where N is smaller than M, determining a related block of the current image block according to a motion vector of the target neighboring block, the current image block, and a reference image of the current image block, dividing the current image block and the related block into a plurality of sub-blocks in a same manner, where the sub-blocks in the current image block correspond to sub-blocks in the related block in a one-to-one correspondence, and performing prediction on the sub-blocks in the current image block according to the motion vectors of the corresponding sub-blocks in the related block, respectively. The complexity of the encoding and decoding can be reduced while the performance of the encoding and decoding is ensured.