Affine Merge Motion Vector Derivation Using Non-Adjacent Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards face challenges in efficiently deriving accurate motion vectors for affine merge mode, leading to increased data requirements for representing motion information, which affects compression efficiency.
Innovation Solution
The method involves obtaining motion vector candidates from non-adjacent neighbor blocks through various scanning strategies and using control point motion vectors to enhance the derivation process for affine motion prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion vector candidates are obtained from adjacent neighbor blocks only, then the derivation process is simple, but the accuracy of motion vector derivation is insufficient
Solution Approach 1:
The patent segments the neighbor block selection into multiple categories: adjacent neighbor blocks, non-adjacent neighbor blocks, and blocks at different scanning distances. This segmentation allows the system to explore a broader candidate pool while maintaining organized derivation processes, thereby improving motion vector accuracy without overwhelming complexity.
Solution Approach 2:
The patent introduces a new dimension to the candidate derivation by incorporating non-adjacent neighbor blocks and multiple scanning distances beyond the traditional adjacent block approach. This dimensional expansion enriches the candidate pool and improves derivation accuracy while managing complexity through structured scanning protocols.
2Measurement precision
If more scanning areas and distances are used to obtain MV candidates, then the accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent implements partial action by allowing flexible selection of scanning areas and distances based on availability and need. The encoder/decoder can adaptively choose to scan only necessary areas or stop when sufficient candidates are found, balancing accuracy improvement with computational complexity management.
Solution Approach 2:
The patent enables dynamic adjustment of scanning parameters such as scanning distance, scanning area, and candidate selection criteria. These parameter changes allow the system to optimize between accuracy and computational complexity based on specific coding conditions, resource availability, and performance requirements.
3Productivity
If traditional affine merge candidate derivation is used, then the data requirements are lower, but the compression efficiency is reduced
Solution Approach 1:
The patent uses copying by deriving motion vector candidates from existing neighbor block motion information rather than performing full motion estimation. This copying approach from adjacent and non-adjacent neighbors reduces the data required for representing motion information while maintaining improved compression efficiency through better candidate accuracy.
Data Source
AI summary
A method of video decoding, a method of video encoding, apparatuses and non-transitory computer-readable storage media thereof are provided. The method of video decoding includes obtaining one or more constructed motion vector (MV) candidates from a plurality of non-adjacent neighbor blocks to a current block by jointly or independently scanning a plurality of scanning areas at one or more scanning distances, where one scanning distance indicates a number of blocks away from one side of the current block. Furthermore, the method may include obtaining one or more control point motion vectors (CPMVs) for the current block based on the one or more constructed MV candidates.