Video Decoder Subblock Motion Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face challenges in efficiently reducing redundancy in video signals, particularly in inter-picture prediction with motion compensation, where existing methods struggle to accurately refine motion vectors and compensate for affine models, leading to suboptimal compression ratios and increased bandwidth requirements.
Innovation Solution
The proposed solution involves processing circuitry configured to decode video bitstreams, refine motion vectors, and apply spatial gradients for improved prediction refinement, using techniques such as subblock-based affine motion models and optical flow to enhance prediction accuracy and reduce data requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If motion compensation with MV prediction is used, then compression ratio is improved, but prediction accuracy deteriorates
Solution Approach 1:
The current block is divided into multiple subblocks, and each subblock is processed independently with its own motion vector refinement. This segmentation allows for more precise local motion compensation while maintaining overall compression efficiency, as each subblock can be optimized separately rather than using a single block-level MV prediction.
Solution Approach 2:
Different refinement processes are applied to different subblocks based on their specific characteristics. The gradient calculation and MV refinement are performed locally for each subblock, allowing the prediction accuracy to be optimized for local motion patterns rather than applying a uniform approach across the entire block.
2Measurement precision
If subblock-based affine motion models are applied, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The block is segmented into subblocks, and affine motion models are applied at the subblock level rather than the entire block level. This reduces the number of parameters that need to be calculated and stored compared to applying a full affine model to the whole block, thereby reducing computational complexity while maintaining prediction accuracy.
Solution Approach 2:
The affine motion model parameters are refined iteratively using gradient calculations. By updating parameters incrementally based on local gradient information, the method achieves high prediction accuracy without requiring excessively complex computational processes, balancing accuracy and complexity through adaptive parameter optimization.
3Measurement precision
If motion vector refinement with gradient calculation is performed, then prediction accuracy is improved, but processing time increases
Solution Approach 1:
The gradient calculation and MV refinement are performed on smaller subblocks rather than the entire block. This segmentation reduces the number of gradient calculations required, as each subblock processes a smaller portion of the image data independently, thereby reducing total processing time while maintaining refinement accuracy for each local region.
Solution Approach 2:
The refinement process applies gradient calculation selectively to subblocks that benefit most from refinement, rather than uniformly processing all blocks at maximum detail. This partial application of the refinement process reduces overall processing time while maintaining sufficient accuracy for the most critical regions.
Data Source
AI summary
A method for video decoding in a decoder is provided. Coding information of a current block (CB) from a coded video bitstream is decoded. The coding information includes weighted prediction information that indicates a weighted prediction for the CB. A determination is made as to whether to apply a prediction refinement with optical flow (PROF) on the CB based on the weighted prediction information. The CB is reconstructed based on the weighted prediction and whether the PROF is determined to be applied on the CB.