Laser Scanner Registration via Clustered Scan Pairing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for registering multiple scans in laser scanning and measurement processes often result in interruptions due to performance limitations, leading to incomplete or ambiguous registrations, especially when not all scans are examined pairwise.
Innovation Solution
A method that uses a position-tracking device to track the laser scanner's positions and register scans into clusters based on quality criteria, forming pairs of scans and comparing their intersections with a threshold value to ensure successful registration, thereby improving the registration process and reducing interruptions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all scans are examined pairwise for registration, then registration accuracy is improved, but processing time and computational complexity increase significantly
Solution Approach 1:
The patent divides the scan set into clusters based on temporal proximity and spatial overlap characteristics. Scans are grouped into clusters where scans within each cluster are more likely to have sufficient overlap for successful registration. This segmentation reduces the total number of pairwise comparisons needed while maintaining registration accuracy within each cluster.
Solution Approach 2:
The patent performs preliminary clustering of scans before the actual registration process. By pre-grouping scans into clusters based on their temporal and spatial relationships, the system prepares the data structure in advance to enable efficient pairwise comparison only within relevant clusters, avoiding unnecessary comparisons across all scans.
2Productivity
If only neighborhood scans are examined for registration, then processing performance is improved, but registration completeness and reliability deteriorate
Solution Approach 1:
The patent introduces a temporal dimension to the clustering process by organizing scans into clusters based on both spatial overlap and temporal proximity. This additional dimension allows the system to efficiently identify relevant scan pairs without exhaustive comparison, improving both performance and reliability by focusing on temporally adjacent scans that are more likely to overlap.
3Device complexity
If clusters are formed without position tracking, then device complexity is reduced, but registration accuracy and scene continuity deteriorate
Solution Approach 1:
The patent introduces scanner position information as an intermediary parameter that mediates between the simple clustering approach and accurate registration. By incorporating position data into the clustering criterion, the system uses this intermediate information to guide scan pairing decisions, improving registration accuracy without requiring complex real-time position tracking integration.
Data Source
AI summary
A method for optically scanning and measuring a scene by a laser scanner includes generating multiple scans; tracking scanner positions with a position-tracking device for the multiple scans and providing tracked scanner positions in response; registering sequentially scans selected from the multiple scans into clusters using registration points or targets and confirming registration of the scans into the clusters according to specified quality criteria being fulfilled; selecting scans from the clusters and forming pairs of scans; forming an intersection of the selected pairs and comparing a size of the intersection with a threshold value obtained based at least in part on the tracked scanner positions; and attempting to register the pairs of scans if the size of the intersection exceeds the threshold value and accepting the registered pairs of scans if the registration is successful.