Binary Tree Splitting for Video Coding Boundary Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In block-based hybrid video coding, forced quad-tree partitioning at frame boundaries can lead to inefficient coding and increased complexity due to the creation of many small coding units, which complicates processing and signaling.
Innovation Solution
The proposed solution involves using binary tree splitting at least in the deepest partitioning hierarchy level or at other levels where coding units are not entirely within the boundary, allowing for coding units of different sizes in horizontal and vertical directions, and optionally combining this with quad-tree splitting in shallower levels, to efficiently handle boundary portions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If forced quad-tree partitioning is applied at frame boundaries, then boundary portions can be processed, but the number of small coding units increases leading to increased complexity
Solution Approach 1:
The patent segments the partitioning process into two distinct phases: forced quad-tree partitioning for initial boundary handling, and optional binary tree partitioning for deeper subdivision. This segmentation allows the system to manage boundary portions systematically while controlling the creation of overly small coding units through the stopping condition at depth-3.
Solution Approach 2:
The patent introduces dynamic control over the partitioning depth through a configurable parameter (depth-3 stopping condition). This allows the system to adapt the partitioning granularity based on specific boundary conditions and coding requirements, transforming a static forced partitioning scheme into a dynamic, controllable process that balances completeness with complexity.
2Manufacturing precision
If forced quad-tree partitioning is applied to create smaller coding units, then boundary alignment is improved, but signaling requirements and processing complexity increase
Solution Approach 1:
The patent extracts the boundary alignment function from the general partitioning process by implementing forced partitioning specifically for boundary portions. This allows precise boundary alignment to be achieved through a dedicated mechanism that operates independently from the standard quad-tree partitioning, reducing the need for complex signaling throughout the entire coding structure.
Solution Approach 2:
The patent applies partial forced partitioning only where necessary (at boundary portions) rather than throughout the entire image. By limiting the forced partitioning to depth-3 and only where boundary alignment is required, the system achieves sufficient boundary precision without the excessive signaling and processing complexity that would result from universal deep partitioning.
3Device complexity
If binary tree splitting is used at deepest level, then number of small coding units is reduced, but partitioning flexibility is limited
Solution Approach 1:
The patent segments the partitioning strategies by applying quad-tree partitioning for the first three depth levels (providing flexibility and structure) and then switching to binary tree partitioning at the deepest level (reducing unit count). This segmentation of partitioning methods across different depth levels allows the system to benefit from both approaches: flexibility where needed and reduced complexity where optimal.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
The present invention relates to partitioning of an image or an image slice. In particular, the image or image slice is split into an integer number of coding tree units in both vertical and horizontal directions. The remaining boundary portion is then at least partly forced-split so that the deepest partitioning level is a result of binary partitioning, i.e. includes a coding unit which is longer on one side than on the other side. The longer side is parallel or concurrent to the image or image slice boundary.