Diagonal Triangle Partitioning for Complex Motion Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently encoding and decoding high-definition and ultra-high-definition video data while maintaining image quality, particularly in the context of evolving video services that require improved coding efficiency.
Innovation Solution
Implementing triangular prediction units in video coding, where video pictures are partitioned into coding units that include triangular-shaped prediction units, with orientation flags and motion vector candidate lists to enhance prediction accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional rectangular block partitioning is used for video coding, then the encoding process is simple and computationally efficient, but prediction accuracy deteriorates for regions with complex motion patterns or diagonal boundaries
Solution Approach 1:
The patent divides a rectangular coding unit into multiple triangular prediction units by introducing a diagonal partition line from one corner to the opposite corner. This segmentation allows different motion characteristics to be modeled in different triangular regions, improving prediction accuracy for complex motion patterns while maintaining a relatively simple overall structure.
Solution Approach 2:
The patent introduces asymmetric triangular partitioning where the coding unit is divided into two unequal triangular regions based on diagonal orientation. This asymmetric division better matches natural image structures and motion patterns that often follow diagonal directions, thereby improving prediction accuracy without requiring complex multi-layer partitioning.
2Productivity
If triangular prediction units are introduced to improve prediction accuracy, then coding efficiency improves, but the complexity of motion vector prediction and candidate list management increases
Solution Approach 1:
The patent creates a universal motion vector candidate list structure that works for both rectangular and triangular prediction units. The same candidate generation rules and selection processes are applied regardless of partition shape, which improves coding efficiency by providing consistent performance while avoiding the need for separate complex prediction mechanisms for different partition types.
Solution Approach 2:
The patent introduces a partition orientation parameter that indicates whether the triangular partition follows one diagonal direction or the other. This single parameter change allows the system to adapt to different triangular configurations without fundamentally changing the motion vector prediction algorithm, thereby improving coding efficiency while controlling complexity through parameter-based adaptation rather than structural redesign.
3Manufacturing precision
If advanced partitioning techniques are used to reduce redundancy, then video quality improves, but the bit rate overhead for signaling partition information increases
Solution Approach 1:
The patent applies triangular partitioning selectively rather than universally - only in regions where complex motion patterns or diagonal boundaries are detected. This partial application reduces the bit rate overhead for signaling partition information while still achieving improved video quality in the specific regions that benefit from the advanced partitioning, avoiding unnecessary overhead in simple regions.
Data Source
AI summary
A method for video coding is provided. The method includes: partitioning video pictures into a plurality of coding units (CUs), at least one of which is further portioned into two prediction units (PUs) including at least one triangular shaped PU with a partitioning orientation in one of: from top-left corner to bottom-right corner, and from top-right corner to bottom-left corner; constructing a uni-prediction motion vector candidate list; determining whether a current CU is coded as triangle prediction mode according to coded information; signaling a partition orientation flag indicating the partitioning orientation; and signaling index values that indicate selected entries in the constructed uni-prediction motion vector candidate list.