3D Scan Pose Registration Using Uncoded Marker Triangles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for stitching 3D scans require manual intervention to assign consistent identities to uncoded markers, leading to inaccuracies and inefficiencies in generating a holistic 3D model.
Innovation Solution
A method and system that automatically determine absolute poses of a scanner by registering 3D scans based on geometric arrangements of uncoded markers, using congruent triangles to find relative poses and generate a holistic 3D representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If uncoded markers are used for stitching 3D scans, then the markers are invariant to scanner viewpoint and can be automatically recognized, but it is not possible to automatically and reliably assign consistent identities to the markers across different scans
Solution Approach 1:
The patent segments the marker identification process into two independent stages: first identifying geometric arrangements (shapes) of markers in each scan, then matching these arrangements across scans to establish consistent identities. This segmentation allows automatic recognition of marker positions while enabling reliable identity assignment through geometric pattern matching, resolving the contradiction between recognition precision and automation extent.
Solution Approach 2:
The patent introduces geometric arrangements (such as triangles formed by marker triplets) as an intermediary representation. Instead of directly matching individual markers across scans, the system first identifies geometric patterns within each scan, then matches these patterns to establish marker identities. This intermediary geometric representation enables automatic and reliable identity assignment that was previously impossible with uncoded markers alone.
2Reliability
If manual intervention is used to assign marker identities, then consistent identities can be assigned across scans, but the process becomes inaccurate, error-prone, and cumbersome
Solution Approach 1:
The patent enables the system to automatically perform marker identity assignment through self-service mechanisms. By using geometric arrangements as intermediaries, the system autonomously identifies and matches markers across scans without human intervention. The geometric pattern matching process automatically establishes consistent identities, making the system self-sufficient and eliminating the need for cumbersome manual operations while maintaining high reliability.
3Extent of automation
If geometric arrangements of markers are used for scan registration, then automatic stitching can be achieved, but multiple pairwise registration processes are required which increases computational complexity
Solution Approach 1:
The patent merges multiple pairwise registration processes into a unified registration framework. By identifying geometric arrangements in each scan and matching these arrangements across all scans simultaneously, the system combines what would otherwise be separate pairwise operations into a single integrated process. This merging reduces computational complexity while maintaining automatic stitching capability, as the geometric pattern matching naturally handles multiple scans without requiring explicit pairwise processing.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The invention relates to reproducing absolute poses of a scanner when capturing a plurality of 3D scans F(s) of an object which is equipped with dedicated markers visible in the scans. In each scan of a pair of scans, triangles formed from representations of the markers in those scans are compared. The best relative pose of the respective scans of the pair is derived from relative poses of congruent triangles of those scans. The best relative poses of all possible scan pairs are then processed to determine the absolute poses.