Video Processing Motion Vector Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding techniques, such as MMVD, face limitations in coding efficiency and quality due to restricted motion directions and fixed distance sets that do not adapt to block size or motion vector magnitude, and lack utilization of neighboring block template information for refinement decisions.
Innovation Solution
The proposed method enhances MMVD by allowing additional diagonal and asymmetric directions, dynamic distance offsets, and context-dependent coding of flags, enabling more flexible and efficient motion vector difference representation through context derivation and adaptive coding strategies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional video coding techniques with fixed motion directions and distance sets are used, then device complexity is reduced, but coding efficiency and quality deteriorate
Solution Approach 1:
The patent applies dynamics by making the motion vector difference (MVD) direction and distance adaptive rather than fixed. The encoder dynamically selects from multiple diagonal and asymmetric directions based on the actual motion characteristics of the current block, and adjusts distance offsets according to block size and motion vector magnitude. This dynamic adaptation improves coding efficiency while managing complexity through context-based derivation.
Solution Approach 2:
The patent changes parameters by introducing variable motion directions (including diagonal and asymmetric angles) and dynamic distance offsets that depend on block size and motion vector magnitude. Instead of using a fixed set of motion directions and distances, the system adjusts these parameters adaptively based on the specific characteristics of each video block, thereby improving coding efficiency and quality.
2Ease of operation
If fixed distance sets are used for MMVD, then ease of operation is improved, but manufacturing precision deteriorates
Solution Approach 1:
The patent changes the distance parameter from a fixed set to a dynamic offset system. The distance offset is adjusted based on the block size and motion vector magnitude, allowing the system to maintain coding simplicity while achieving higher motion vector precision. This parameter adaptation enables more accurate motion representation without significantly increasing operational complexity.
3Measurement precision
If neighboring block template information is not utilized, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent applies feedback by utilizing neighboring block template information to guide the selection of motion vector difference directions and distances. The encoder uses the motion characteristics of neighboring blocks as feedback to refine the current block's motion parameters, improving measurement precision. This feedback mechanism is implemented efficiently through context-based derivation that leverages already-decoded neighboring information.
4Adaptability or versatility
If restricted motion directions are used, then device complexity is reduced, but adaptability deteriorates
Solution Approach 1:
The patent makes the motion direction selection dynamic by introducing multiple diagonal and asymmetric directions that adapt to the specific motion patterns in the video content. Instead of being restricted to a fixed set of directions, the system can select from a broader range of angles based on the actual motion characteristics, thereby improving adaptability while managing complexity through structured derivation methods.
Solution Approach 2:
The patent applies asymmetry by introducing asymmetric motion directions that better match real-world motion patterns. Rather than relying solely on symmetric horizontal and vertical directions, the system incorporates diagonal and asymmetric angles that more accurately represent complex motion trajectories, enhancing adaptability without proportionally increasing structural complexity.
Data Source
AI summary
Embodiments of the present disclosure provide a solution for video processing. A method for video processing is proposed. The method comprises: performing a conversion between a current video block of a video and a bitstream of the video based on at least one flag of: a first flag indicating whether a merge mode with motion vector difference (MMVD) is used for the current video block, or a second flag indicating whether an affine MMVD is used for the current video block, wherein the at least one flag is bypass coded or is coded with at least one context determined from a plurality of contexts. Thereby, the proposed method can advantageously improve coding efficiency and coding quality.