Affine Motion Vector Correction for Video Encoding Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding methods face challenges in efficiently encoding and decoding video data while maintaining high image quality, due to limited bandwidth and the need to balance compression rate and image quality, particularly in inter prediction processes where accuracy is compromised with larger data unit sizes.
Innovation Solution
The implementation of a video encoding and decoding method using an affine mode and ultimate motion vector expression (UMVE) mode, which involves obtaining base motion vectors from adjacent blocks, correcting them with specific information, and determining affine motion vectors to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If the size of data units is increased to reduce encoding information, then the amount of encoding information is decreased, but prediction accuracy is reduced
Solution Approach 1:
The patent divides the current block into multiple sub-blocks and performs separate affine motion compensation for each sub-block. This segmentation allows the use of smaller effective block sizes for prediction while still working with larger data units, thereby maintaining prediction accuracy without excessively increasing encoding information.
Solution Approach 2:
The patent changes the parameter of motion vector representation by introducing correction information (delta values) that refines base motion vectors. This allows more precise motion description without linearly increasing the amount of encoding information, as the correction values are typically smaller than full motion vectors.
2Measurement precision
If the number of prediction methods is increased to improve prediction accuracy, then prediction accuracy is improved, but the amount of encoding information is increased
Solution Approach 1:
The patent employs a dynamic selection mechanism where the affine prediction mode is conditionally applied based on block characteristics and motion complexity. The encoder decides whether to use affine mode with sub-block partitioning or conventional methods, allowing adaptive optimization without committing to a fixed large set of prediction methods for all blocks.
Solution Approach 2:
The patent applies affine prediction with sub-block partitioning selectively to regions where motion complexity justifies the additional encoding overhead. By making the prediction method local and adaptive rather than universal, the system improves accuracy where needed while minimizing unnecessary encoding information in simpler regions.
Data Source
AI summary
A video decoding method includes: obtaining two or more base motion vectors from an adjacent block of a current block; obtaining correction information for correcting the two or more base motion vectors; determining two or more affine motion vectors by correcting the two or more base motion vectors according to the correction information; obtaining a plurality of affine parameters of the current block according to the two or more affine motion vectors; and predicting the current block according to the plurality of affine parameters.