Multi-Model Syntax Design for Video Coding Block Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face inefficiencies in block partitioning, particularly with the increasing complexity of combining quadtree and binary tree structures, which affects coding efficiency and computational complexity, especially in advanced methods like Multi-Type-Tree (MTT) block partitioning.
Innovation Solution
A multi-model syntax design is introduced to reduce signalling overhead for CU structure syntax, allowing different syntax models to be used based on CU structure characteristics, such as leaf CU sizes and partitioning depths, to efficiently encode or decode block regions using quadtree and 1D partitions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If quadtree and binary tree structures are combined for block partitioning, then partitioning flexibility is improved, but device complexity increases
Solution Approach 1:
The patent segments the block partitioning process into two distinct stages: first applying quadtree partitioning to divide the current block into sub-blocks, then applying binary tree partitioning to further divide selected sub-blocks. This segmentation allows the system to achieve high partitioning flexibility through multiple partition types while managing complexity by processing different partition types in sequence rather than simultaneously
Solution Approach 2:
The patent implements dynamic partitioning by allowing the binary tree partitioning to be conditionally applied based on the results of quadtree partitioning. The system dynamically determines which sub-blocks require further partitioning and applies binary tree splitting only where needed, rather than applying a fixed partitioning structure throughout. This dynamic approach maintains flexibility while reducing unnecessary computational complexity
2Productivity
If multi-model syntax design is used for CU partitioning, then coding efficiency is improved, but signalling overhead increases
Solution Approach 1:
The patent applies partial action by implementing multi-model syntax design only for the necessary CU partitioning scenarios. Rather than using complex syntax models for all possible partitioning cases, the system selectively applies different syntax models based on the specific partitioning situation. This partial application of complex syntax reduces signalling overhead while maintaining coding efficiency where it is most needed
Solution Approach 2:
The patent changes the syntax parameters dynamically based on the partitioning stage and type. Different syntax elements and parameter sets are used for quadtree partitioning versus binary tree partitioning. This parameter adaptation allows the system to optimize coding efficiency for each partitioning type while minimizing the total signalling overhead by only transmitting necessary parameters for the actual partitioning performed
Data Source
AI summary
A method and apparatus of video coding for a video encoding system or video decoding system are disclosed. According to one method, the current block region is partitioned into one or more leaf blocks using quadtree (QT) partition and/or 1D (one-dimensional) partition by firstly signalling or parsing a first syntax element to indicate whether the current block region is split into multiple smaller blocks. According to a second method, a target syntax model from multiple syntax models is determined. The current block region is partitioned into one or more leaf blocks using QT partition and/or 1D partition and signalling block partition information according to the target syntax model at an encoder side, or the block partition information is parsed according to the target syntax model and the current block region is partitioned into one or more leaf blocks at a decoder side.