Video Coding Partition Tree Restrictions via VPDU Size
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards face limitations in partition tree restrictions, particularly with the assumption of whole-block transforms and common test conditions, which may lead to issues such as dependent enabling or disabling of splitting based on maximally allowed transform sizes, and inadequate handling of sub-block transform boundaries.
Innovation Solution
The proposed solution involves determining whether a partition tree is allowed based on the virtual pipeline data unit (VPDU) size and maximally allowed sub-block transform sizes, allowing or disallowing splitting accordingly, and applying filtering processes such as deblocking filters based on sub-block transform boundaries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If partition tree splits are allowed based on maximally allowed transform sizes, then video coding flexibility is improved, but handling of sub-block transform boundaries becomes inadequate
Solution Approach 1:
The patent changes the parameter used for determining partition tree split validity from maximally allowed transform size to VPDU size. This parameter change allows the partition tree to be restricted based on the actual pipeline data unit boundaries, thereby improving both the flexibility of partitioning and the precision of boundary handling. The VPDU size serves as a more appropriate constraint that aligns with the actual processing units in the video coding pipeline.
2Device complexity
If whole-block transform assumptions are used, then partition tree restrictions are simplified, but sub-block transform boundary handling is inadequate
Solution Approach 1:
The patent applies local quality by making the partition tree restriction rule dependent on the specific location and size of the VPDU. Instead of using a global maximally allowed transform size assumption, the restriction is locally adapted to each VPDU's dimensions. This allows the system to maintain simple restriction logic while improving boundary handling precision, as each partition tree split is evaluated against the actual VPDU boundaries relevant to that specific block.
3Ease of manufacture
If maximally allowed transform sizes are used for partition restrictions, then implementation is easier, but video coding efficiency is reduced
Solution Approach 1:
The patent changes the implementation parameter from abstract maximally allowed transform sizes to concrete VPDU sizes that are already defined in the video coding standard. This parameter change maintains ease of implementation because VPDU size is an existing parameter in the coding framework, while simultaneously improving video coding efficiency by ensuring partition tree splits align with actual processing boundaries, reducing unnecessary computations and improving overall coding performance.
Data Source
AI summary
Devices, systems and methods for digital video coding. A method for video processing includes determining whether to apply and how to apply a filtering process for samples of a current video unit of a video based on whether the samples are located at predefined transform boundaries; and performing a conversion between the video and a bitstream representation of the video based on the determining, where the predefined transform boundaries are determined based on a first transform mode in which a transform region is split into multiple transform sub-regions and only one transform sub-region has a non-zero residual sample.