Point Cloud Attribute Compression via KD Tree Intra-Frame Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point cloud attribute compression methods face challenges in balancing compression performance with computational complexity, with methods like octree and KD tree decomposition and graph transformation either underutilizing spatial correlation or having high operation complexity.
Innovation Solution
An intra-frame prediction-based point cloud attribute compression method using a new block structure-based scheme with four prediction modes, employing KD tree division, color space transformation, and DCT transformation to reduce information redundancy among coding blocks, optimizing compression performance while maintaining low calculation complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If octree decomposition and DCT are used for point cloud attribute compression, then calculation complexity is low, but spatial correlation between points is not fully utilized and compression performance needs improvement
Solution Approach 1:
The patent divides the point cloud into multiple coding blocks using KD tree decomposition, creating a hierarchical structure that enables localized processing. This segmentation allows the method to balance computational efficiency with compression performance by operating on smaller, manageable blocks while preserving spatial relationships within each block.
Solution Approach 2:
The patent introduces graph transformation as an additional processing dimension beyond traditional DCT. By constructing graphs where nodes represent points and edges represent spatial relationships, the method adds a topological dimension to the compression process, enabling better exploitation of spatial correlation without significantly increasing computational complexity.
2Loss of information
If octree decomposition and graph transformation are used for point cloud attribute compression, then compression performance is good, but operation complexity is high and subgraph problems occur
Solution Approach 1:
The patent segments the point cloud into coding blocks at multiple levels of the KD tree hierarchy. This multi-level segmentation reduces the size of graphs that need to be constructed and processed, thereby reducing operation complexity and avoiding subgraph problems while maintaining good compression performance through hierarchical processing.
Solution Approach 2:
The patent dynamically adjusts the graph construction and processing based on the specific characteristics of each coding block. By adaptively determining graph parameters and processing strategies for different blocks, the method optimizes the balance between compression performance and operation complexity for each local region.
3Loss of information
If KD tree decomposition and graph transformation are used for point cloud attribute compression, then the subgraph problem is solved and compression performance is greatly improved, but operation complexity is still high
Solution Approach 1:
The patent employs multi-level KD tree decomposition to divide the point cloud into hierarchically organized coding blocks. This segmentation strategy reduces the scale of graphs requiring transformation, thereby lowering operation complexity while maintaining the compression performance benefits of graph transformation through progressive processing at different decomposition levels.
Solution Approach 2:
The patent applies graph transformation selectively to certain coding blocks rather than uniformly to all blocks. By identifying and applying graph transformation only where it provides significant compression benefits, the method reduces overall operation complexity while maintaining high compression performance in critical regions.
Data Source
AI summary
An intra-frame prediction-based point cloud attribute compression method. A new block structure-based intra-frame prediction scheme is provided for point cloud attribute information, where four prediction modes are provided to reduce information redundancy among different coding blocks as much as possible and improve point cloud attribute compression performance. The method comprises: performing point cloud attribute color space conversion; dividing a point cloud by using a K-dimensional (KD) tree to obtain coding blocks; performing block structure-based intra-frame prediction; performing intra-frame prediction mode division; performing conversion, uniform quantization, and entropy coding. The method relates to performing coding block division on a point cloud by using a KD tree, numbering coding blocks in a breadth-first traversal order, processing the coding blocks one by one in the numbering order by using a point cloud intra-frame prediction scheme having four prediction modes, selecting an optical prediction mode by means of mode decision, and performing conversion, quantization, and entropy coding on a predicted residual, so as to achieve optimal performance of point cloud attribute compression.