Point Cloud Registration via Geometric Segmentation and Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for aligning multiple laser scan point cloud data sets fail to accurately merge scans taken from different vantage points, resulting in incomplete coverage of areas like buildings, as each scan lacks coordinate alignment, making it difficult to determine the full measurements of structures.
Innovation Solution
The solution involves segmenting imported scans based on geometric attributes, filtering segments for registration, and using geometric primitives like planar segments to align them in a geometric manner, employing a registration process that includes coarse and fine refinement steps to determine proper alignment and create a unified coordinate space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If multiple laser scans are taken from different vantage points to capture large structures, then coverage area is improved, but coordinate alignment accuracy deteriorates
Solution Approach 1:
The patent segments point cloud data into multiple planar segments representing different surfaces (walls, floors, ceilings). Each scan is divided into geometric primitives that can be independently matched across scans, enabling accurate coordinate alignment while maintaining comprehensive coverage of large structures
Solution Approach 2:
The patent uses planar segments as intermediary elements to bridge multiple scans. These segments serve as common reference features that exist across different vantage points, allowing the system to establish coordinate transformations between scans through matching segment geometries rather than direct point-to-point alignment
2Measurement precision
If all scan segments are processed through registration, then alignment accuracy is improved, but computational time deteriorates
Solution Approach 1:
The patent segments point cloud data into multiple planar segments representing different surfaces (walls, floors, ceilings). Each scan is divided into geometric primitives that can be independently matched across scans, enabling accurate coordinate alignment while maintaining comprehensive coverage of large structures
Solution Approach 2:
The patent extracts only the essential geometric features (planar segments with specific attributes) from the full point cloud data for registration processing. By taking out only the relevant planar segments rather than processing all points, the system achieves accurate alignment while significantly reducing computational time
Data Source
AI summary
A method, system, apparatus, article of manufacture, and computer-readable storage medium provide the ability to merge multiple point cloud scans. A first raw scan file and a second raw scan file (each including multiple points) are imported. The scan files are segmented by extracting segments based on geometry in the scene. The segments are filtered to reduce a number of segments and identify features. A set of candidate matching feature pairs are acquired by coarsely registering features from one scan to features from another scan. The candidate pairs are refined by improving alignment based on corresponding points in the features. The candidate pairs are scored and then merged based on the scores.