MMVD Motion Vector Offset Selection for Video Decoding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards face challenges in efficiently encoding and decoding high-definition and ultra-high-definition video data while maintaining image quality, particularly in the use of merge mode with motion vector differences (MMVD) tools.
Innovation Solution
Implementing methods and apparatus to enhance the coding efficiency of MMVD by receiving and applying motion vector difference (MVD) offsets and inter prediction filter (InterPF) modes based on control flags, allowing for more precise motion vector reconstruction in video data decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If merge mode with motion vector differences is used, then coding efficiency is improved, but motion vector precision and prediction accuracy are limited
Solution Approach 1:
The patent applies dynamics by making the motion vector difference (MVD) offsets adaptive rather than fixed. The decoder dynamically selects from multiple predefined MVD offset sets based on the actual motion characteristics of the video content, allowing the system to adjust its precision level according to the specific coding situation while maintaining high coding efficiency
Solution Approach 2:
The patent changes the parameter of motion vector precision by introducing multiple sets of predefined MVD offsets with different precision levels. Instead of using a single fixed precision, the system can switch between different offset sets (e.g., quarter-pel, half-pel, one-pel precision) to optimize the balance between coding efficiency and prediction accuracy for different video content types
2Ease of manufacture
If fixed MVD offsets are used, then implementation is simple, but prediction accuracy is insufficient for varying video content
Solution Approach 1:
The system transitions from a static, fixed MVD offset implementation to a dynamic one where the decoder can adaptively select from multiple predefined offset sets. This maintains implementation simplicity by using predefined tables while enabling accurate adaptation to varying video content characteristics through runtime selection based on motion information
Solution Approach 2:
The patent introduces parameter variability by providing multiple sets of predefined MVD offsets with different precision characteristics. The system can select the appropriate offset set based on the video content's motion properties, thereby improving prediction accuracy without requiring complex real-time calculation, thus balancing simplicity and accuracy
3Measurement precision
If higher motion vector precision is applied, then prediction accuracy is improved, but coding complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-defining multiple sets of MVD offsets with different precision levels before the actual coding process. This allows the system to prepare multiple precision options in advance and simply select the appropriate one during decoding, avoiding the need for complex real-time precision adjustment while still achieving high prediction accuracy when needed
Solution Approach 2:
The system manages coding complexity by providing multiple predefined parameter sets (MVD offset sets) rather than requiring continuous or adaptive parameter calculation. The decoder simply needs to select from pre-computed options based on motion characteristics, which maintains high precision for complex scenes while keeping the overall coding process manageable and not excessively complex
Data Source
AI summary
An electronic apparatus performs a method of decoding video data. The method comprises: receiving, from a bitstream, a first control flag that indicates merge mode with motion vector difference (MMVD) is enabled for one or more coding units in a video sequence; receiving a first syntax from the video data that identifies a set of motion vector difference (MVD) offsets from a plurality sets of MVD offsets; receiving, a second control flag corresponding to a respective coding unit of the one or more coding units, which indicates the MMVD is applied to the coding unit; receiving a second syntax that selects an MVD offset from the identified set of MVD offsets, and a third syntax that selects an MVD direction; forming MVD based on the selected MVD offset and MVD direction; and reconstructing the coding unit by applying the formed MVD to generate motion vectors to the coding unit.