Boundary CTU Multi-Type Splitting for Flexible Video Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face limitations in the flexibility and efficiency of partitioning video frames, particularly at image boundaries, leading to suboptimal compression and decoding performance.
Innovation Solution
The proposed solution involves hierarchical multi-type splitting of coding tree units (CTUs) with adaptable and predefined maximum partition depths, allowing for enhanced flexibility in partitioning both non-boundary and boundary CTUs, using a combination of quad-tree and multi-type tree structures to optimize partitioning depth.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a fixed maximum partition depth is applied to all coding tree units, then the device complexity is reduced and processing is simplified, but the adaptability and compression efficiency at image boundaries deteriorate
Solution Approach 1:
The patent applies different maximum partition depths to different spatial locations: non-boundary CTUs use a first maximum depth while boundary CTUs use a second, greater maximum depth. This local differentiation allows the system to maintain simpler processing in regular areas while achieving better compression at boundaries where it is most needed.
Solution Approach 2:
The patent introduces dynamic adaptability by allowing the maximum partition depth to vary based on the CTU's location relative to image boundaries. The system dynamically selects between a first maximum depth for non-boundary CTUs and a second maximum depth for boundary CTUs, enabling flexible adaptation to different coding scenarios.
2Productivity
If deeper partitioning is applied to boundary CTUs, then compression efficiency and decoding performance improve, but the device complexity and processing overhead increase
Solution Approach 1:
The patent concentrates the increased partitioning depth specifically at boundary CTUs where it provides the most compression benefit, while maintaining the shallower first maximum depth for non-boundary CTUs. This localized approach ensures that the complexity increase is confined to only the regions where it is most beneficial for compression efficiency.
Solution Approach 2:
The patent changes the partition depth parameter based on CTU location: using a first maximum depth for non-boundary CTUs and a second, greater maximum depth for boundary CTUs. This parameter adaptation allows the system to optimize compression efficiency at boundaries without uniformly increasing complexity across the entire image.
3Adaptability or versatility
If multi-type splitting is used with extended maximum depth for boundary CTUs, then the adaptability and compression performance improve, but the processing time and computational load increase
Solution Approach 1:
The patent applies the computationally intensive multi-type splitting with extended maximum depth only to boundary CTUs, while using the simpler first maximum depth for non-boundary CTUs. This localized application reduces the overall processing time by limiting the complex operations to only the regions where they provide the most value.
Solution Approach 2:
The patent dynamically adjusts the partitioning strategy based on CTU location, switching between a simpler mode for non-boundary CTUs and a more flexible multi-type splitting mode for boundary CTUs. This dynamic adaptation allows the system to achieve high adaptability at boundaries while maintaining reasonable processing speeds overall.
Data Source
AI summary
The present disclosure provides apparatuses and methods for splitting an image into coding units. An image is divided into coding tree units (CTUs) which are hierarchically partitioned. Hierarchical partitioning includes multi-type partitioning such as binary tree or quad tree splitting. For CTUs completely within the image and CTUs on the boundary, respective multi-type partition depths are chosen. The present disclosure provides for multi-type partitioning flexibility in a boundary portion of the image.