Point Cloud Registration Methods for Long-Span Bridge Spatial Deformation Monitoring Using Terrestrial Laser Scanning
Hongwei Zhang,Yanjie Zhu,Wen Xiong,C. S. Cai
DOI: https://doi.org/10.1155/2023/2629418
2023-02-09
Structural Control and Health Monitoring
Abstract:In recent years, efforts have been devoted to utilizing terrestrial laser scanning for bridge spatial performance inspection, but they are still restricted to small or medium-span bridges, like some historical heritages. Due to the large-scale dimensional features of long-span bridges, applications of 3D point cloud techniques still remain challenging, such as the extra-long scan range and extreme-small incidence angle when scanning a bridge with a span over 1000 m. Moreover, rare attempts can be found for the performance evaluation of point cloud registration methods for long-span bridges as well, which is a critical basis for further spatial deformation recognition on the point cloud data. Hence, in this study, a cross-evaluation of three iterative closest point (ICP) registration methods is conducted for long-span suspension bridges, namely, traditional ICP, kd-tree-based ICP, and feature point-based ICP algorithms. We conducted field laser scanning on the Ma’anshan Yangtze Bridge, an 1880 m long suspension bridge located in China. The results show that the feature point-based ICP algorithm outperforms the other two in terms of convergence rate and execution time for a single iteration due to the smaller number of registration points compared to the other two algorithms. Moreover, it also gives more precise values in terms of bridge tower spatial deformation identification. Meanwhile, due to efficient data organization, the kd-tree-based ICP algorithm takes less time for a single iteration than the traditional ICP algorithm. Finally, two suggestions for algorithm improvement in terms of efficiency optimization and accuracy improvement of long-span bridge deformation analysis are proposed based on the assessment results.
construction & building technology,engineering, civil,instruments & instrumentation