Video Decoding Motion Vector Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards face inefficiencies in motion vector prediction, particularly in refining motion vectors for inter prediction, which affects coding efficiency and computational complexity during video decoding and encoding.
Innovation Solution
A video decoding and encoding device with a processing circuit that determines candidate motion vectors, refines them using diamond or cross search patterns, and selects the refined motion vector based on a selection index, optimizing motion vector refinement for improved coding efficiency and reduced complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If motion vector refinement is applied to all candidate motion vectors, then coding efficiency is improved, but computational complexity increases
Solution Approach 1:
The patent changes the parameter of refinement application from universal to selective based on motion vector characteristics. Specifically, refinement is applied only when the motion vector difference exceeds a threshold or under specific prediction modes, transforming the processing approach from exhaustive to conditional, thereby reducing computational load while maintaining coding efficiency for critical cases
Solution Approach 2:
The patent applies partial action by refining only a subset of candidate motion vectors rather than all of them. The refinement process is selectively applied based on predetermined conditions such as motion vector difference thresholds or specific prediction mode requirements, achieving sufficient coding performance without the excessive computational cost of universal refinement
2Reliability
If motion vector refinement is applied to multiple identical candidate motion vectors, then coding efficiency is improved, but redundancy increases computational overhead
Solution Approach 1:
The patent extracts and processes only the necessary candidate motion vectors for refinement by identifying and eliminating duplicates beforehand. When multiple identical candidate motion vectors exist, the system selects one representative for refinement and excludes others, removing redundant processing while maintaining the beneficial refinement effect
Solution Approach 2:
The patent discards redundant identical candidate motion vectors after selecting one representative for refinement. The refinement result from the selected vector is then recovered and applied, avoiding repeated refinement of identical vectors while preserving the coding efficiency benefits
Data Source
AI summary
Aspects of the disclosure include a video decoding device that includes a processing circuit. The processing circuit is configured to receive encoded video data associated with a current block of a current image frame, determine a set of candidate motion vectors for the current block, determine a selection index based on the encoded video data, and select one of the set of candidate motion vectors based on the selection index. The processing circuit is further configured to refine the selected candidate motion vector to become a reconstructed motion vector when the selected candidate motion vector meets a predetermined condition, and generate an image of the current image frame for output, including generating a reconstructed image of the current block based on the reconstructed motion vector.