Sub-Pixel Motion Vector Refinement for FRUC Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding technologies, particularly in HEVC, face inefficiencies in motion vector prediction due to sub-pixel refinement processes that do not fully leverage signal characteristics and matching types, leading to suboptimal coding efficiency.
Innovation Solution
Implement a decoder-side motion vector derivation method that refines motion vectors at sub-pixel positions, using adaptive search patterns and resolutions based on signal content and edge directions, to improve accuracy and reduce bitrates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sub-pixel motion vector refinement is performed using conventional methods, then motion vector accuracy is improved, but coding efficiency is suboptimal due to not leveraging signal characteristics
Solution Approach 1:
The patent applies local quality by adapting the refinement process to local signal characteristics. Different refinement strategies are used depending on the matching type (uni-prediction vs bi-prediction) and the presence of edges in templates. The search pattern and resolution are adjusted locally based on template content, allowing optimal refinement for each specific case rather than using a fixed approach.
Solution Approach 2:
The refinement process is made dynamic by adjusting search patterns and resolutions based on real-time analysis of template signals. The system dynamically selects between different search patterns (diamond, cross, horizontal, vertical) and resolutions (1/4, 1/8, 1/16 sub-pixels) depending on the detected edge directions and matching type, optimizing the refinement process for each specific scenario.
2Ease of manufacture
If fixed sub-pixel search patterns are used for motion vector refinement, then processing is simplified, but coding efficiency is limited by not adapting to signal characteristics
Solution Approach 1:
The system transitions from fixed search patterns to dynamic, adaptive search patterns that change based on template analysis. The search pattern is selected according to the matching type and edge directions detected in templates, allowing the process to adapt to different signal characteristics while maintaining systematic refinement.
Solution Approach 2:
The patent changes multiple parameters including search pattern type, search resolution, and search direction based on template characteristics. The resolution can vary between 1/4, 1/8, and 1/16 sub-pixels, and the search pattern can be diamond, cross, horizontal, or vertical depending on the detected edge directions, optimizing each parameter for the specific signal conditions.
3Measurement precision
If higher sub-pixel resolution is used for motion vector refinement, then motion vector accuracy is improved, but processing complexity increases
Solution Approach 1:
The system applies local quality by using higher resolutions (1/8 or 1/16 sub-pixels) only when necessary based on template analysis. When templates show clear edge structures, higher resolution refinement is applied to achieve better accuracy. When templates are simpler, lower resolution (1/4 sub-pixels) is sufficient, reducing unnecessary processing complexity.
Solution Approach 2:
The resolution parameter is made dynamic rather than fixed. The system dynamically adjusts the search resolution between 1/4, 1/8, and 1/16 sub-pixels based on the complexity and characteristics of the templates being processed. This dynamic adjustment ensures high accuracy when needed while avoiding excessive complexity when simpler refinement suffices.
Data Source
AI summary
A method and a device for encoding or decoding video data. It concerns more particularly the encoding according to a particular encoding mode using a decoder side motion vector derivation mode referenced as frame-rate up conversion mode or FRUC mode. In FRUC merge mode, the derivation process comprises a refinement step to increase the accuracy of the obtained motion vector at the sub-pixel level. This process involves the evaluation of different sub-pixel position around the obtained motion vector according to different patterns. The present invention has been devised to improve the known refinement step. It aims at improving the coding efficiency by considering the characteristics of the matching type and/or the signal inside the templates.