Video Signal Motion Vector Refinement for Inter-Prediction Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video signal processing technologies face inefficiencies in inter-prediction and motion compensation, particularly when handling high-resolution and stereographic image content, leading to increased costs in transmission and storage due to the large data volumes involved.
Innovation Solution
A method and apparatus for encoding and decoding video signals that derive an initial motion vector from a merge candidate list, refine it using a motion refinement vector, and combine these to determine the final motion vector, with the option to encode/decode information about the size and direction of the refinement vector.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image compression technology is used for high-resolution images, then image quality is maintained, but transmission and storage costs increase due to large data volumes
Solution Approach 1:
The patent applies parameter changes by refining motion vectors from integer precision to sub-integer precision (1/4 pixel, 1/8 pixel, or 1/16 pixel accuracy). This refinement allows for more precise motion compensation, improving prediction accuracy and reducing residual data that needs to be encoded and transmitted, thereby reducing overall data volume while maintaining or improving image quality
2Productivity
If motion compensation with multiple merge candidates is implemented, then inter-prediction efficiency is improved, but device complexity increases
Solution Approach 1:
The patent segments the motion vector refinement process into distinct stages: selecting an initial motion vector from merge candidates, determining a refinement vector separately, and combining them. This segmentation allows the system to manage complexity by breaking down the refinement process into manageable steps while still achieving high prediction accuracy through multiple candidates
Solution Approach 2:
The patent implements partial refinement by optionally encoding/decoding refinement vector information based on conditions such as block size, motion vector magnitude, or prediction mode. This partial action approach improves efficiency when refinement is beneficial while avoiding unnecessary processing complexity when refinement would provide minimal gain
Data Source
AI summary
According to the present invention, there is provided a method of decoding an image, the method including: deriving an initial motion vector of a current block; determining a motion refinement vector of the current block; and determining a motion vector of the current block on the basis of the initial motion vector and the motion refinement vector. Herein, the initial motion vector is derived from any one of merge candidates included in a merge candidate list for the current block.