Flexible CTU Boundary Partitioning for Video Decoding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In modern video coding standards like H.264/AVC, HEVC, VVC, and AV1, picture partitioning that misaligns with object boundaries requires additional bits to signal boundaries and motion, leading to inefficiencies.
Innovation Solution
The method involves partitioning a coded picture into coding tree units (CTUs) where CTUs adjacent to picture boundaries have reduced dimensions, allowing for flexible decoding by signaling size dimensions and restricting certain splitting modes based on these dimensions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a fixed picture partitioning into CTUs is used, then the decoding structure is simple and regular, but additional bits are required to signal boundary and motion information when object boundaries do not align with CTU boundaries
Solution Approach 1:
The patent introduces flexible CTU row heights that can be dynamically adjusted based on picture content characteristics. Instead of fixed CTU dimensions, the system allows CTU rows to have variable heights, enabling better alignment with object boundaries while maintaining a relatively simple decoding structure. This dynamic adaptation reduces the need for additional boundary and motion information signaling.
Solution Approach 2:
The patent changes the parameter of CTU row height from a fixed value to a variable parameter. By allowing CTU row heights to be adjusted according to the picture content, the system achieves better alignment with object boundaries, thereby reducing the bits required for boundary and motion information while keeping the overall decoding structure manageable.
2Loss of information
If CTUs at picture boundaries are reduced in size, then alignment with object boundaries improves and signaling bits are reduced, but the decoding complexity increases due to variable dimensions
Solution Approach 1:
The patent segments the picture into CTU rows with variable heights, where the first and last CTU rows can have different heights from intermediate rows. This segmentation allows boundary regions to be handled with appropriate sizing while maintaining regularity in the bulk of the picture, thus reducing signaling bits without excessively increasing decoding complexity.
Solution Approach 2:
The patent applies different CTU row height characteristics to different regions of the picture. The first and last CTU rows (at picture boundaries) can have reduced heights to align with object boundaries, while intermediate CTU rows maintain standard heights. This local adaptation reduces boundary information bits while limiting the overall increase in decoding complexity to only the boundary regions.
3Manufacturing precision
If flexible CTU partitioning is implemented, then object boundary alignment improves, but the number of syntax elements and signaling requirements increase
Solution Approach 1:
The patent implements dynamic CTU row height selection where the height of first and last CTU rows can be flexibly determined based on picture content. This dynamic approach improves object boundary alignment precision while using syntax elements only where necessary (at boundaries), rather than requiring comprehensive syntax elements for the entire picture, thus limiting the increase in signaling complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and methods may provide flexible picture partitioning, a method includes receiving a coded picture that is partitioned into a plurality of coding tree units (CTUs), wherein at least one row or column of CTUs, among the plurality of CTUs of the coded picture, that is adjacent to a boundary of the coded picture has a size dimension that is smaller than a corresponding size dimension of each CTU among the plurality of CTUs that is not adjacent to any boundary of the coded picture; and decoding the coded picture based on the plurality of CTUs, wherein the at least one row or column of CTUs includes a first CTU row or a first CTU column of the coded picture that is adjacent to a top boundary or left boundary of the coded picture, respectively.