Point Cloud Attribute Encoding Using Lossless Residuals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current point cloud encoding methods face inaccuracies in prediction, leading to reduced effectiveness in encoding and decoding processes due to inefficient compression of attribute information, particularly in handling large datasets.
Innovation Solution
A method that determines prediction values for attribute information based on geometry information, processes residual values with lossless encoding, and reconstructs attribute information using prediction and residual values to improve encoding accuracy without significant impact on bitstream size.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If prediction is performed using neighboring points' attribute information, then compression of attribute information is achieved, but prediction accuracy is insufficient leading to reduced encoding effectiveness
Solution Approach 1:
The patent changes the parameter of prediction by introducing prediction offset values that adjust the predicted attribute information. Instead of directly using neighboring points' attribute information as prediction values, the system calculates offset values based on geometry information and applies them to refine the prediction, thereby improving prediction accuracy while maintaining compression effectiveness.
Solution Approach 2:
The patent replaces the simple mechanical copying of neighboring attribute values with a more sophisticated system that incorporates geometry information. By substituting the direct copying mechanism with a prediction mechanism that uses geometric relationships and offset calculations, the system achieves higher prediction accuracy without sacrificing compression ratios.
2Manufacturing precision
If lossless encoding is performed on residual values, then encoding accuracy is improved, but bitstream size increases
Solution Approach 1:
The patent applies local quality by performing lossless encoding selectively on specific residual values rather than uniformly on all residual values. The system identifies which residual values require lossless encoding based on their magnitude or significance, applying lossless encoding only where necessary to maintain encoding accuracy, while allowing lossy encoding for less critical values to control bitstream size.
Solution Approach 2:
The patent implements partial action by applying lossless encoding to only a subset of residual values rather than all residual values. This selective approach ensures that the most important attribute information is preserved with high accuracy while avoiding the bitstream size increase that would result from applying lossless encoding universally.
Data Source
AI summary
A point cloud encoding and decoding method and a point cloud decoder are provided in the disclosure. The point cloud encoding includes the following. Geometry information and attribute information of points in a point cloud are obtained. Prediction values of the attribute information of the points in the point cloud are determined according to the geometry information of the points in the point cloud. Residual values of the attribute information of the points in the point cloud are determined according to the prediction values of the attribute information of the points in the point cloud. The residual values of the attribute information of the points in the point cloud are processed with a first encoding process, where lossless encoding is performed on a residual value of attribute information of at least one point in the point cloud in the first encoding process.