Decoder Side Motion Vector Refinement Subblock Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards, such as HEVC and VVC, face limitations in achieving optimal motion vector refinement, particularly in decoder-side processing, which affects compression efficiency and accuracy, especially when dealing with complex motion patterns and varying block sizes.
Innovation Solution
The proposed method involves a multi-level decoder side motion vector refinement (DMVR) process that splits coding blocks into subblocks, allowing for refined motion vector generation and update at multiple levels, and a template-based DMVR approach that relaxes symmetric offset constraints, enabling more flexible and accurate motion vector prediction across various coding modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single-level DMVR process is performed on the entire coding block, then the processing complexity is low, but the motion vector prediction accuracy is insufficient for complex motion patterns
Solution Approach 1:
The coding block is divided into multiple subblocks, and DMVR is performed separately on each subblock. This segmentation allows the refinement process to capture local motion variations within different regions of the coding block, significantly improving motion vector prediction accuracy for complex motion patterns while keeping the complexity manageable through localized processing.
Solution Approach 2:
Different motion vector refinement strategies are applied to different subblocks based on their local characteristics. The DMVR process is adapted to handle varying motion complexities in different regions, allowing high accuracy where needed while reducing unnecessary processing in simpler regions, thus balancing accuracy and complexity.
2Adaptability or versatility
If symmetric offset constraints are enforced in DMVR, then the processing is simpler and more consistent, but the adaptability to different coding modes and motion patterns is reduced
Solution Approach 1:
The DMVR process dynamically adjusts the offset constraints based on the coding mode and motion characteristics. Instead of enforcing fixed symmetric constraints, the algorithm adapts the refinement strategy to match the specific coding scenario, enabling better versatility across different coding modes while maintaining reasonable processing complexity through adaptive decision-making.
3Productivity
If motion vector refinement is performed at multiple levels on subblocks, then the compression efficiency is improved, but the computational complexity increases
Solution Approach 1:
The coding block is divided into multiple subblocks for independent DMVR processing. This segmentation enables parallel computation of motion vectors for different regions, which improves compression efficiency by capturing local motion details while managing computational complexity through distributed processing that can be implemented efficiently in hardware or through parallel computation architectures.
Data Source
AI summary
The present disclosure provides systems and methods for processing video content. The method can include: generating, for a coding block, a motion vector (MV) in a first coding mode; and updating the MV by performing a decoder side motion vector refinement (DMVR) process on the coding block.