A Robust Measurement Method for Geometric Profiles of Bridges Based on Prior Design Knowledge Orientation

Abstract Terrestrial laser scanning (TLS) technology is increasingly used to measure bridge geometric shapes. However, the commonly observed loss of critical geometric details in point clouds poses significant challenges for automate and accurate measurement. A robust measurement method based on prior design knowledge orientation is proposed in the present study. First, component point cloud slices are extracted from full-bridge point cloud data through manual segmentation and 3D-to-2D projection. Then, the prior design knowledge is generated from the bridge design data, including component cross-sectional point clouds and geometric information. This prior knowledge is integrated with the extracted slices through template matching to establish stable geometric constraints. Finally, the extracted feature points under constraints are connected to generate the component alignment. The proposed method was validated through a field test on an under-construction arch bridge and a robustness evaluation using a virtual scanning bridge model. The results show that the method achieves a measurement accuracy within 5 mm and exhibits significant tolerance to variations in data density and completeness.