Point Cloud Attribute Prediction Using Filter Matrix
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current point cloud attribute compression methods fail to accurately predict attribute values due to neglecting the position relationship between neighbor points, leading to suboptimal prediction accuracy in irregular and unordered point cloud distributions.
Innovation Solution
A point cloud attribute prediction method and device that determines K nearest neighbor points and calculates a filter matrix to improve prediction accuracy by considering both the geometrical relationship between the current point and its neighbors, as well as the relationship among neighbor points, using methods like Euclidean, Manhattan, or geodesic distances for neighbor selection and filter determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional attribute prediction methods are used that only consider the position relationship between neighbor points and the current point, then the prediction process is simple, but the prediction accuracy is insufficient due to neglecting the position relationship among neighbor points themselves
Solution Approach 1:
The patent introduces a filter matrix as an intermediary tool that captures the position relationships among neighbor points. This filter matrix serves as a mediator between the neighbor points and the current point, enabling the prediction process to consider both the relationship between neighbors and the current point, as well as the relationships among neighbors themselves, thereby improving prediction accuracy without excessively complicating the process
Solution Approach 2:
The patent changes the parameters of the prediction process by incorporating the filter matrix with different kernel functions (Gaussian, exponential, rational quadratic). By changing these parameters and selecting appropriate kernel functions, the system can adapt to different point cloud distributions and improve prediction accuracy while maintaining manageable computational complexity
2Productivity
If K-nearest neighbor algorithm is used to select neighbor points, then neighbor points can be found efficiently, but it cannot be guaranteed that all neighbor points are uniformly distributed at the periphery of the current point in irregular and unordered point cloud distributions
Solution Approach 1:
The filter matrix acts as an intermediary that compensates for the non-uniform distribution of neighbor points. Even when K-nearest neighbor algorithm selects points that are not uniformly distributed, the filter matrix with appropriate kernel functions can weight and smooth the contributions of these points, effectively capturing the underlying spatial relationships and improving prediction accuracy
Solution Approach 2:
The patent adjusts parameters such as the bandwidth of the kernel function in the filter matrix to adapt to different point cloud densities and distributions. By changing these parameters, the system can optimize the influence radius and weighting of neighbor points, ensuring better performance even when uniform distribution cannot be guaranteed
Data Source
AI summary
Provided are a point cloud attribute prediction method and prediction device based on a filter. The method comprises a coding method and a decoding method, and the device comprises a coding device and a decoding device. The method comprises: determining K nearest neighbor points of the current point; determining a filter matrix; and determining an attribute prediction value of the current point according to the filter matrix. Therefore, the compression performance of a point cloud attribute can be improved by means of selecting an appropriate filter.