Point Cloud Attribute Compression via KD-Tree Segmentation and Adaptive Intra-Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point cloud attribute compression methods face challenges in balancing compression performance with computational complexity, with high operation complexity often resulting in suboptimal efficiency and redundancy issues.
Innovation Solution
A method using a KD tree for dividing point cloud coded blocks and employing six multi-angle self-adaptive intra prediction modes to reduce redundancy, with color space conversion, intra prediction, and optimal mode decision to achieve efficient compression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If graph transformation is used for point cloud attribute compression, then compression performance is improved, but operation complexity increases
Solution Approach 1:
The patent segments the point cloud into multiple transformation blocks using KD-tree decomposition, processing each block independently with graph transformation. This segmentation allows the complex graph transformation operation to be applied locally to smaller blocks rather than the entire point cloud, improving compression performance while managing operational complexity through divide-and-conquer
Solution Approach 2:
The patent applies different processing strategies to different regions by performing graph transformation locally within each transformation block. Each block's attributes are transformed based on its specific characteristics while using spatial correlation with neighboring blocks, achieving local optimization that improves overall compression performance without uniformly increasing complexity across the entire dataset
2Manufacturing precision
If KD tree decomposition with graph transformation is applied, then compression performance is greatly improved, but operation complexity remains high
Solution Approach 1:
The patent combines KD-tree decomposition with graph transformation by segmenting the point cloud into transformation blocks first, then applying graph transformation within each block. This segmentation reduces the operational complexity of graph transformation by limiting its scope to smaller blocks while maintaining the compression performance benefits through localized attribute transformation and spatial correlation exploitation
Solution Approach 2:
The patent dynamically adjusts the transformation block size and graph connectivity parameters based on the local characteristics of each point cloud region. By adapting the processing parameters to local conditions rather than using fixed settings, the method achieves high compression performance while optimizing operational complexity for different data characteristics
3Productivity
If existing JPEG encoder with DCT is used, then calculation complexity is low, but spatial correlation between points is not fully utilized
Solution Approach 1:
The patent introduces graph transformation as an intermediary step between the spatial decomposition and the final encoding. The graph transformation acts as a mediator that exploits spatial correlation by transforming attributes based on neighborhood relationships, preparing the data in a way that enhances compression performance while maintaining compatibility with standard encoders like JPEG that use DCT
Solution Approach 2:
The patent creates a composite compression approach by combining multiple techniques: KD-tree decomposition for spatial organization, graph transformation for exploiting spatial correlation, and DCT-based encoding for final compression. This composite method integrates the advantages of each technique, achieving both good compression performance and reasonable computational efficiency
Data Source
AI summary
Disclosed is a multi-angle adaptive intra-frame prediction-based point cloud attribute compression method. A novel block structure-based intra-frame prediction scheme is provided for point cloud attribute information, where six prediction modes are provided to reduce information redundancy among different coding blocks and improve the point cloud attribute compression performance. The method comprises: (1) inputting a point cloud; (2) performing point cloud attribute color space conversion; (3) dividing a point cloud by using a K-dimensional (KD) tree to obtain coding blocks; (4) performing block structure-based multi-angle adaptive intra-frame prediction; (5) performing intra-frame prediction mode decision; and (6) performing conversion, uniform quantization, and entropy encoding.