Decoder-Side Motion Refinement Using Extended Sample Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding techniques face challenges in achieving high coding efficiency while minimizing intensity discontinuities at sub-block boundaries and maintaining low computational complexity and memory requirements, especially for small or feature-less blocks.
Innovation Solution
The method involves extending the block size used for computing the dis-similarity or similarity measure during motion vector refinement by including predicted samples within a certain horizontal and vertical distance from the closest boundary samples of candidate predicted blocks, thereby improving the coherence of refined motion vectors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the block size is extended for matching, then the quality of match is improved, but the internal memory requirements increase
Solution Approach 1:
The patent divides the extended block into multiple regions: a central region containing the original block and boundary samples from neighboring sub-blocks, and optional outer regions extending further. This segmentation allows the matching process to access a larger effective area while managing memory usage through structured organization of the extended samples.
Solution Approach 2:
The patent extends the matching block in spatial dimensions by including boundary samples from adjacent sub-blocks horizontally and vertically. This dimensional extension allows the matching process to consider a broader context without requiring a complete increase in block size, thereby improving match quality while controlling memory requirements through targeted extension.
2Productivity
If Motion Vector Refinement is performed on non-overlapping sub-blocks, then processing efficiency is improved, but intensity discontinuities occur at sub-block boundaries
Solution Approach 1:
The patent merges boundary samples from multiple neighboring sub-blocks into a unified extended block for matching. By combining samples from different sub-block boundaries into a single extended matching region, the method ensures intensity continuity across sub-block boundaries while maintaining the processing efficiency of independent sub-block refinement.
Solution Approach 2:
The extended block acts as an intermediary structure that includes boundary samples from multiple sub-blocks. This intermediary allows the matching process to bridge intensity discontinuities at sub-block boundaries by considering context from all adjacent regions, thereby maintaining coherence while preserving processing efficiency.
3Measurement precision
If the block size is increased for small or feature-less blocks, then matching quality is improved, but computational complexity increases
Solution Approach 1:
The patent applies local quality by extending the block size specifically at boundary regions where neighboring sub-block samples are included, rather than uniformly increasing the size of all blocks. This localized extension improves matching quality for small or feature-less blocks while minimizing the increase in computational complexity by only expanding where boundary context is needed.
Data Source
AI summary
A method for inter prediction of a block of a picture and corresponding apparatus are provided. The method comprises: determining a first prediction block based on an initial first motion vector of a current block in a first reference picture, so as to obtain a first extension block; obtaining from the first reference picture, one or more first reference blocks each of which is associated with a combination of the initial first motion vector and one of motion vector offsets; obtaining a target motion vector offset with a smallest cost among the motion vector offsets; determining for the current block, a refined first motion vector which is a combination of the initial first motion vector and the target motion vector offset; and obtaining a refined prediction block of the current block based on the refined first motion vector, thereby improving coherence of refined motion vectors to achieve higher coding efficiencies.