Point Cloud Matching via Feature Segmentation and Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining the matching relationship between point cloud data, such as the Iterative Closest Point (ICP) algorithm, require extensive computation and labor, are inefficient for large-scale in-vehicle laser point cloud data with non-rigid deformation characteristics, and consume significant memory and time, failing to guarantee accurate matching.
Innovation Solution
A method that classifies points in a point cloud into categories like ground, planar, and columnar points, generates new characteristics from these categories, and determines the matching relationship by comparing these characteristics, reducing computational requirements and allowing registration of partial degrees of freedom instead of 6 degrees, thereby improving speed and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the ICP algorithm is used to determine the matching relationship between point cloud data, then the matching accuracy is improved, but the computational time and memory space consumption increase significantly
Solution Approach 1:
The patent segments the point cloud data into multiple local regions and performs matching operations on each region separately. This divides the large-scale computational problem into smaller, more manageable sub-problems, reducing the overall computational time and memory space requirements while maintaining matching accuracy through localized analysis.
Solution Approach 2:
The patent extracts key geometric features and characteristics from the point cloud data, such as surface normals, curvature information, and local geometric descriptors. By extracting only the essential features needed for matching rather than processing all raw point data, the computational load is significantly reduced while preserving the ability to achieve accurate matching results.
2Measurement precision
If manual extraction of characteristic points is performed to determine matching relationship, then the matching precision is improved, but the labor cost and data requirements increase
Solution Approach 1:
The patent implements automated feature extraction algorithms that automatically identify and extract characteristic points from the point cloud data without requiring manual intervention. The system uses geometric analysis and local feature detection to self-identify key matching points, thereby eliminating labor costs while maintaining high matching precision through algorithmic feature selection.
3Productivity
If conventional methods are used for in-vehicle laser point cloud with non-rigid deformation, then the processing speed is maintained, but the matching accuracy deteriorates
Solution Approach 1:
The patent introduces dynamic adaptation mechanisms that allow the matching algorithm to adjust to non-rigid deformations in real-time. The system dynamically modifies matching parameters and feature selection based on the detected deformation characteristics of the point cloud data, enabling accurate matching even when the object undergoes non-rigid transformations while maintaining processing speed through efficient algorithmic updates.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Embodiments of the present disclosure provide a method, apparatus and computer readable storage medium for determining a matching relationship between point cloud data. The method includes extracting a first characteristic associated with first point cloud data and a second characteristic associated with second point cloud data. The first point cloud data and the second point cloud data are acquired for the same object. The method further includes performing characteristic matching between the first characteristic and the second characteristic. In addition, the method further includes determining, based on the characteristic matching, a matching relationship between the first point cloud data and the second point cloud data.