QTBT Video Block Partitioning for Higher Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current HEVC video coding systems rely solely on quadtree partitioning for CTU, CU, and PU, which limits coding efficiency due to restrictive partition types and lack of adaptability to varying video content characteristics.
Innovation Solution
The implementation of a binary tree partitioning process that splits blocks into two primary sub-blocks of equal or different sizes, combined with quadtree partitioning to generate leaf nodes, allowing for more flexible and adaptive block partitioning structures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If quadtree partitioning is used to partition CTU into CUs, then the partitioning process is simple and systematic, but the adaptability to local characteristics is limited and coding efficiency is constrained
Solution Approach 1:
The patent divides the block partitioning process into two distinct stages: quadtree partitioning (dividing a block into four equal sub-blocks) and binary tree partitioning (dividing a block into two unequal sub-blocks). This segmentation allows the system to first apply simple quadtree partitioning for coarse segmentation, then apply binary tree partitioning for finer adaptation to local characteristics, thereby resolving the contradiction between simplicity and adaptability.
Solution Approach 2:
The patent introduces dynamic switching between quadtree and binary tree partitioning modes based on local block characteristics. The encoder can adaptively select which partitioning mode to apply at each node of the coding tree, allowing the partitioning structure to dynamically adjust to varying local characteristics rather than following a fixed quadtree-only pattern.
2Productivity
If quadtree partitioning is applied recursively to partition blocks, then the partitioning structure is regular and easy to implement, but coding efficiency and BD-rate performance are limited
Solution Approach 1:
The patent merges quadtree partitioning and binary tree partitioning into a unified Coding Tree Structure (CTS). This combination allows the system to leverage the regularity and simplicity of quadtree partitioning while incorporating the flexibility and coding efficiency of binary tree partitioning, achieving improved BD-rate performance without excessive complexity.
Solution Approach 2:
The patent applies different partitioning strategies to different regions of the image based on local characteristics. Binary tree partitioning is applied selectively to regions where it provides coding efficiency benefits, while quadtree partitioning is used in other regions, ensuring that each local area is processed with the most appropriate partitioning method for its specific characteristics.
3Manufacturing precision
If minimum CU size is set to 8×8 for HEVC, then the granularity for switching prediction types is fine, but the partitioning process becomes more complex and coding efficiency is reduced
Solution Approach 1:
The patent adds a new dimension to the partitioning process by introducing binary tree partitioning as a second partitioning mode alongside the traditional quadtree. This allows the system to achieve fine prediction switching granularity through selective application of binary tree partitioning at deeper levels, without requiring all blocks to be partitioned to the minimum 8×8 size, thereby maintaining coding efficiency.
Data Source
AI summary
A method and apparatus of video coding using block partitioning process including a binary tree partitioning process are disclosed. The block partitioning structure corresponding to the block partitioning process for a block of video data is derived by parsing a first indicator associated with a current block and determining whether the binary tree partitioning process applies a binary tree partition to the current block based on a value of the first indicator. When it is determined that the binary tree partition is applied to the current block, the current block is always split into two primary sub-blocks with same size or two primary sub-blocks with different sizes. The block partitioning structure represents partitioning the block of video data into final sub-blocks. The block of video data is decoded based on the final sub-blocks decoded according to the block partitioning structure derived.