Template-Guided Hierarchical Multi-View Registration Framework of Unordered Bridge Terrestrial Laser Scanning Data

Area of Science:

• Geomatics Engineering
• Civil Engineering
• Computer Vision

Background:

• Bridge point cloud data (PCD) registration is crucial for modeling, deformation detection, and health monitoring.
• Existing multi-view registration methods are inefficient due to extensive pairwise matching and graph construction.
• A need exists for improved methods to handle unordered multiple scans effectively.

Purpose of the Study:

• To propose a marker-free, template-guided method for aligning multiple unordered bridge PCDs to a global coordinate system.
• To enhance the efficiency and accuracy of multi-view bridge PCD registration.
• To recover scan overlaps and establish a merging order without extensive pairwise matching.

Main Methods:

• A template-guided approach aligns individual scans to a registration template, recovering overlaps.
• A graph-partition algorithm constructs scan-blocks from overlaps and scan locations.
• Coarse-to-fine registration within blocks and global block-to-block registration achieve unified alignment.

Main Results:

• The proposed method successfully aligns multiple unordered bridge PCDs to a global coordinate system.
• High accuracy was demonstrated on datasets from suspension and continuous rigid frame bridges.
• The framework efficiently recovers overlaps and establishes a merging order for scans.

Conclusions:

• The marker-free template-guided method offers a highly accurate and efficient solution for multi-view bridge PCD registration.
• This approach addresses limitations of existing pairwise registration techniques.
• The method is suitable for various bridge types, supporting advanced structural analysis and monitoring.