3D Point Cloud Registration Using Stable Normal-Vector Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D scanners require time-consuming manual registration processes and post-processing steps to align multiple scans, necessitating user interaction and the use of external control networks or artificial targets, which hampers efficiency in scanning large areas.
Innovation Solution
A method for automatically registering 3D point clouds by identifying a subset of stable, high-quality data points with consistent normal vectors, allowing for efficient alignment of scans without manual intervention or external targets, using a 3D laser scanner with integrated image acquisition and distance measurement technologies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual registration procedures are used to align multiple scans, then registration accuracy can be improved, but the process becomes time-consuming and requires user interaction
Solution Approach 1:
The system performs automatic self-registration by identifying stable points and computing transformation parameters autonomously without requiring manual user intervention or external control networks, thereby eliminating time-consuming manual operations while maintaining registration accuracy
Solution Approach 2:
The invention extracts and utilizes naturally occurring stable points (such as corners and edges) from the scanned environment itself as registration features, eliminating the need for external artificial targets or control networks, thus reducing both time and operational complexity
2Reliability
If external control networks or artificial targets are used for registration, then registration reliability can be improved, but device complexity and operational requirements increase
Solution Approach 1:
The laser scanner is designed to perform both primary scanning functions and registration functions using the same device and naturally occurring features, eliminating the need for separate external control networks or artificial targets, thereby reducing device complexity while maintaining registration reliability
Solution Approach 2:
The system uses the scanned environment's own geometric features (corners, edges) as registration targets, making the system self-sufficient and eliminating dependencies on external equipment or人工 targets, thus simplifying the overall system while ensuring reliable registration
3Area of stationary object
If all data points are used for registration, then comprehensive coverage is improved, but processing time increases due to inclusion of unstable points
Solution Approach 1:
The system applies different quality criteria to different data points, identifying and selecting only stable points (those with consistent normal vectors across multiple scans) for registration processing, while excluding unstable points, thereby maintaining comprehensive coverage of the scanned area while reducing processing time through selective optimization
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables rapid, automated registration of dynamic and unstructured environments, reducing the need for user interaction and external targets, and enhancing scanning efficiency by focusing on stable points for accurate alignment.
Implementation Method 1
A TOF laser scanner is a scanner in which the distance to a target point is determined based on the speed of light in air between the scanner and a target point
Implementation Method 2
The beam steering mechanism includes a first motor that steers the beam of light about a first axis by a first angle that is measured by a first angular encoder
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of registering three-dimensional (3D) point clouds may include obtaining a first 3D point cloud acquired at a first location; obtaining a second 3D point cloud acquired at a second location; calculating a first normal vector for each point of the first 3D point cloud to create a plurality of normal vectors; calculating, for each point of the first 3D point cloud, a normal deviation amount of the corresponding normal vector to other normal vectors in a predetermined neighborhood of the point; selecting, from the first 3D point cloud, a first registration region based on whether the normal deviation amount of each point meets a deviation threshold; and registering the first 3D point cloud and the second 3D point cloud to create the composite 3D point cloud, the registration utilizing the first registration region in place of the first 3D point cloud.