Shifted CTU Grid Alignment for Video Compression Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression systems, such as HEVC and VVC, have limited flexibility in representing Coding-Tree Units (CTUs) which leads to inefficient compression due to fixed grid alignment, resulting in an increased number of small CUs in high spatial activity areas.
Innovation Solution
The method involves shifting the CTU grid by determining shifting offsets to align boundaries with spatial activity differences in the video picture, allowing CTUs to cover homogeneous areas, thereby reducing the number of CUs and improving compression efficiency without increasing computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the CTU grid is aligned with fixed boundaries, then the encoding structure is simple and regular, but the compression efficiency is reduced due to increased number of small CUs in high spatial activity areas
Solution Approach 1:
The patent changes the position parameter of the CTU grid by introducing shifting offsets (ctbLeftOffset and ctbTopOffset) that allow the grid to be shifted horizontally and vertically. This parameter change enables the CTU boundaries to align with spatial activity boundaries in the video picture, reducing the number of small CUs in high activity areas and improving compression efficiency without fundamentally changing the grid structure
2Productivity
If the CTU grid is shifted to align with spatial activity boundaries, then the number of CUs is reduced and compression efficiency improves, but the grid alignment flexibility increases complexity
Solution Approach 1:
The patent applies local quality by allowing different regions of the video picture to have different CTU alignment characteristics. The shifting offsets are determined based on local spatial activity boundaries, enabling the CTU grid to adapt locally to content characteristics while maintaining a consistent global grid structure. This localized adaptation improves encoding efficiency without requiring complete grid reconfiguration
3Device complexity
If fixed size CTUs are used throughout the picture, then the encoding process is simple, but high spatial activity areas generate more small CUs reducing compression performance
Solution Approach 1:
The patent introduces dynamic adjustment to the CTU grid position by calculating shifting offsets based on spatial activity analysis of the video picture. The grid remains fixed in size but its position becomes dynamic, adapting to the content characteristics. This dynamic positioning reduces the number of small CUs in high activity areas while maintaining the simplicity of fixed-size CTU partitioning
Data Source
AI summary
A method for encoding a video picture into a bitstream of encoded video picture data, includes: obtaining a grid of coding-tree units to split at least one component of the video picture into coding-tree units, each coding-tree unit being a picture area subdivided according to a coding tree; determining at least one shifting offset by aligning at least one boundary of the grid of coding-tree units with at least one boundary separating picture areas with low spatial activity from picture areas with high spatial activity of the video picture; shifting the grid of coding-tree units according to the at least one shifting offset; obtaining encoded video picture data by encoding at least one coding unit (CU) of a coding tree associated with each coding-tree unit (CTU) of the shifted grid of coding-tree units; and writing the encoded video picture data into the bitstream.