Adaptive Motion Vector Resolution for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression techniques face challenges in achieving improved compression ratios with minimal sacrifice in image quality, especially when dealing with limited network resources and increasing demands for higher video quality.
Innovation Solution
The method involves determining and signaling separate horizontal and vertical motion vector resolutions (MVRs) in a bitstream to balance prediction accuracy and compression efficiency, allowing for optimized precision in both horizontal and vertical components of motion vectors, which are then used to position blocks in video frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single motion vector resolution is used for both horizontal and vertical components, then the encoding complexity is reduced, but the coding efficiency and video quality are compromised due to inability to optimize independently in different directions
Solution Approach 1:
The motion vector resolution parameter is segmented into separate horizontal and vertical components. Instead of using a single resolution value for both dimensions, the patent independently determines and encodes horizontal MVR and vertical MVR, allowing each direction to be optimized separately for better coding efficiency while maintaining manageable encoding complexity through modular processing.
Solution Approach 2:
The patent applies different motion vector resolutions to different spatial directions (horizontal and vertical) based on local optimization needs. By allowing independent selection of MVR for each direction, the system achieves local quality optimization where each dimension can use the most appropriate resolution for its specific requirements, improving overall video quality without uniformly increasing complexity.
2Measurement precision
If higher motion vector precision is used, then the prediction accuracy and video quality are improved, but the bandwidth requirements and data transmission costs increase
Solution Approach 1:
The patent dynamically changes the motion vector resolution parameter based on content characteristics and direction. By allowing independent selection of horizontal and vertical MVR values, the system can adjust the precision parameter to match the actual motion patterns in different directions, achieving high prediction accuracy where needed while reducing precision (and thus bandwidth) where lower precision suffices.
3Manufacturing precision
If motion vector resolution is increased to improve video quality, then the image quality is enhanced, but the compression ratio deteriorates due to increased data size
Solution Approach 1:
The patent applies different motion vector resolutions to different spatial directions based on local optimization needs. By allowing independent selection of MVR for each direction, the system achieves local quality optimization where each dimension can use the most appropriate resolution for its specific requirements, improving overall video quality without uniformly increasing complexity.
Solution Approach 2:
The motion vector resolution is made dynamic and adaptive rather than fixed. The patent allows the MVR to vary independently in horizontal and vertical directions based on content characteristics, enabling the system to optimize the balance between video quality and compression ratio dynamically rather than using a static resolution setting.
Data Source
AI summary
A mechanism of video coding is provided. The mechanism includes determining a motion vector predictor (MVP), a horizontal motion vector difference component (MVDx), and a vertical motion vector difference component (MVDy) associated with a motion vector (MV). A horizontal MV resolution (MVR) and a vertical MVR are also determined for the MV. A horizontal component of the MVP is rounded to a precision specified by the horizontal MVR and a vertical component of the MVP is rounded to a precision specified by the vertical MVR. The MV is determined based on the MVP, the MVDx, and the MVDy. The MV is then employed to position a current block in a current frame relative to a reference block in a reference frame. A video stream including the current frame is generated for display on a screen.