Multi-type-tree video coding partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding techniques struggle with efficient block partitioning, particularly in capturing objects centered within blocks, leading to suboptimal coding performance.
Innovation Solution
The implementation of a multi-type-tree (MTT) framework for partitioning video blocks, which allows for the use of three or more different partition structures at each depth of a tree structure, enabling more flexible and efficient coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional block partitioning methods are used, then the encoding process is simple, but the coding efficiency is insufficient and objects in the center of blocks cannot be captured effectively
Solution Approach 1:
The patent applies segmentation by dividing the block partitioning process into multiple independent tree structures (first, second, and third partitioning trees). Each tree handles specific partitioning orientations and types, allowing the system to selectively apply appropriate partitioning strategies to different regions of the video block. This segmentation enables more precise capture of objects in the center of blocks while maintaining manageable complexity through modular organization.
Solution Approach 2:
The patent implements nesting by organizing multiple partitioning trees within a unified multi-type-tree framework. The first, second, and third partitioning trees are nested within the overall block partitioning structure, with each tree containing nodes that can be further subdivided. This nested structure allows hierarchical processing where coarser partitioning decisions at upper levels guide finer partitioning at lower levels, improving coding efficiency without overwhelming complexity.
2Manufacturing precision
If multiple partition structures are used to capture objects in the center of blocks, then coding efficiency improves, but the complexity of the partitioning system increases
Solution Approach 1:
The patent applies local quality by assigning different partitioning characteristics to different regions and orientations. The first partitioning tree handles horizontal partitions, the second handles vertical partitions, and the third handles diagonal partitions. Each tree node can be independently configured with specific partitioning parameters, allowing the system to apply precise partitioning strategies locally to capture objects in the center of blocks while keeping the overall system organized and manageable.
Solution Approach 2:
The patent implements dynamics by making the partitioning structure adaptive and configurable. The multi-type-tree framework allows dynamic selection of which partitioning tree to use based on the local content characteristics. Nodes within each tree can be dynamically subdivided or left as leaves depending on the encoding requirements. This dynamic behavior enables the system to achieve high partitioning precision for capturing centered objects while avoiding unnecessary complexity in regions where simple partitioning suffices.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of decoding video data including receiving a bitstream that includes a sequence of bits that forms a representation of a coded picture of the video data, partitioning the coded picture of the video data into a plurality of blocks using three or more different partition structures, and reconstructing the plurality of blocks of the coded picture of the video data. Partitioning the coded picture of the video data may include partitioning the coded picture of the video data into the plurality of blocks using the three or more different partition structures, wherein at least three of the three or more different partition structures may be used at each depth of a tree structure that represents how a particular block of the coded picture of the video data is partitioned.