Voxel-Based 3D Scan Alignment for Marker-Free Large Object Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for aligning 3D scan datasets face challenges such as precise alignment without references, lack of dimension verification, and scalability constraints, particularly when dealing with large objects like airplanes.
Innovation Solution
A computing system that generates a voxel-based representation of a target object, computes transformation matrices for scan point sets using a fit error minimization operation, and applies these matrices to align scan points within an outer surface voxel layer, utilizing a parameter space search algorithm to minimize alignment errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If external tracking systems or markers are used for alignment, then alignment precision is improved, but device complexity and operational cost increase
Solution Approach 1:
The system uses the scanned object itself to provide alignment references through voxel map generation. The object's own geometric features are extracted and used as registration targets, eliminating the need for external tracking systems or markers. This self-referential approach maintains high alignment precision while reducing device complexity and operational cost.
Solution Approach 2:
The system creates a voxel-based digital copy of the scanned object that serves as a reference model for alignment. This virtual replica contains all necessary geometric information needed for precise registration without requiring physical reference markers on the actual object.
2Measurement precision
If traditional point-to-surface alignment methods are used, then alignment accuracy is improved, but processing speed and scalability deteriorate with large datasets
Solution Approach 1:
The system segments the scanned point cloud data into multiple scan point sets that can be processed independently and in parallel. Each subset is aligned separately to the voxel map, reducing the computational burden on any single processing unit while maintaining overall alignment accuracy across the complete dataset.
Solution Approach 2:
The system replaces traditional iterative point-to-surface alignment algorithms with a voxel-based registration approach. The voxel map provides a discrete, structured representation that enables faster comparison and alignment operations, substituting complex continuous mathematical operations with more efficient discrete spatial indexing and matching.
3Loss of information
If comprehensive scan data from multiple angles is collected, then measurement completeness is improved, but data processing time and computational cost increase
Solution Approach 1:
The system performs preliminary voxel map generation from the complete multi-angle scan data before conducting the actual alignment operations. This pre-processing step organizes all the comprehensive measurement information into an efficient data structure, allowing subsequent alignment operations to proceed rapidly without reprocessing the raw scan data.
Solution Approach 2:
The system transforms the three-dimensional point cloud data into a voxel-based representation that adds a discrete spatial indexing dimension. This voxelization converts continuous spatial coordinates into discrete grid cells, enabling more efficient storage, retrieval, and comparison operations while preserving all geometric information from the multi-angle scans.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A computing system (10) for aligning laser scan data is provided, including a processor (12) configured to obtain multiple scan point sets (22), each set (22) including a plurality of scan points (24) that represent an outer surface of a target object (40). The processor (12) is configured to generate a voxel map (32) including an outer surface voxel layer (36) for the target object (40), and compute a respective transformation matrix (52) for each of the scan point sets (22). The processor (12) is configured to apply the respective transformation matrix (52) for each scan point set (22) to the scan points (24) in the respective scan point set (22) to thereby transform a position of each of the scan points (24) in each scan point set (22) that lies partially or wholly outside of the outer surface voxel layer (36) to a modified position within the outer surface voxel layer (36), to thereby generate respective transformed scan point sets (60) containing the scan points (62) at the modified positions.