Point Cloud Registration Using Pre-learned Feature Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current point cloud registration technologies are inefficient in unifying point cloud data from different view angles, which hampers their application in fields like building protection and 3D mapping.
Innovation Solution
A method that involves obtaining feature vectors of points in point cloud data, determining key points, and matching them across different data sets using pre-created feature models and grid-based search techniques to improve registration efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional point cloud registration methods are used to unify point cloud data from different view angles, then registration accuracy can be maintained, but the processing efficiency is low and the computational complexity is high
Solution Approach 1:
The patent segments the point cloud data processing into distinct stages: key point detection, feature vector extraction, and matching. By dividing the dense point cloud into key points only, the computational complexity is reduced while maintaining registration accuracy. The feature vectors are extracted independently for each key point, enabling parallel processing and improving efficiency.
Solution Approach 2:
The patent extracts only the essential key points and their feature vectors from the complete point cloud data, discarding redundant information. This extraction process reduces the data volume significantly while preserving the critical geometric features needed for accurate registration, thereby improving processing efficiency without sacrificing accuracy.
2Measurement precision
If all points in the point cloud data are processed for registration, then comprehensive matching can be achieved, but the number of points to be learned increases and processing time extends
Solution Approach 1:
The patent performs preliminary key point detection and feature vector extraction before the actual matching process. By pre-processing the point cloud data to identify and extract only the necessary key points and their features, the subsequent matching operation becomes much faster while maintaining precision. This preliminary action eliminates redundant computations on non-key points.
Solution Approach 2:
The patent applies different processing quality to different parts of the point cloud data: key points receive full feature vector extraction and detailed processing, while non-key points are effectively skipped. This local quality approach ensures that computational resources are concentrated on the most important points for registration accuracy, reducing overall processing time without compromising precision.
Data Source
AI summary
A method, an electronic device and a readable storage medium for point cloud data processing, which may be used for autonomous driving, are disclosed. The feature vectors of respective points in the first point cloud data and second point cloud data are pre-learned, and thus the feature vectors of the first key points may be determined directly based on the learnt second feature vectors of respective first neighboring points of the respective first key points in the first point cloud data, and the feature vectors of the candidate key points may be determined directly based on the learnt third feature vectors of the respective second neighboring points of respective candidate key points in the second point cloud data corresponding to the first key points.