Bi-Prediction Weight Indexing for Affine Merge Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution, high-quality images and videos, particularly in virtual reality and augmented reality, necessitates a more efficient image/video compression technique to reduce transmission and storage costs.
Innovation Solution
The method involves deriving weight index information for bi-prediction, using affine merge candidates with motion vectors for control points, and generating reconstructed samples based on weighted averages to enhance image coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional compression techniques are used for high-resolution images and videos, then transmission and storage costs increase, but image quality and resolution requirements cannot be met
Solution Approach 1:
The patent changes the parameters of motion vector prediction by introducing affine transformation models with multiple control points (CP0, CP1, CP2, CP3) at different positions within the block. This allows the prediction to adapt to complex motion patterns in high-resolution videos, improving compression efficiency without sacrificing image quality.
Solution Approach 2:
The patent divides the prediction block into multiple regions by using control points at different positions (top, bottom, left, right edges). Each control point independently predicts motion for its associated region, allowing localized adaptation to different motion characteristics within the same block, thereby improving both compression ratio and visual quality.
2Measurement precision
If affine merge candidates with multiple control points are used, then motion prediction accuracy improves, but computational complexity increases
Solution Approach 1:
The patent performs preliminary construction of affine merge candidate lists by pre-calculating motion vectors for control points at block edges. These candidates are prepared in advance and stored for selection, reducing the computational burden during the actual prediction process while maintaining high motion prediction accuracy.
Solution Approach 2:
The patent implements a selective approach where affine merge candidates are constructed only when necessary (e.g., when regular merge candidates are insufficient). The system can choose to use 2, 3, or 4 control points depending on the complexity of the motion pattern, avoiding unnecessary computational overhead while maintaining accuracy when needed.
3Measurement precision
If weight index information is derived for bi-prediction, then prediction accuracy improves, but encoding complexity increases
Solution Approach 1:
The patent applies different weight indices to different regions of the prediction block by associating weight information with specific control points. Each control point can have its own weight index, allowing localized optimization of prediction accuracy for regions with different motion characteristics, while the overall encoding complexity is managed through selective application.
Data Source
AI summary
According to a disclosure of the present document, when the type of inter-prediction of a current block is indicated as biprediction, weight index information for a candidate within a merge candidate list or a sub-block merge candidate list may be derived, and coding efficiency may be increased.