An adaptive cross-section extraction algorithm for deformation analysis

Abstract

In view of the difficulties in extracting cross-section information and the lack of applicable deformation analysis based on the point cloud, an adaptive cross-section extraction algorithm for deformation analysis is proposed in our study. Firstly, to extract the boundary points along the point cloud route direction, the boundary detection and feature points identification algorithm based on the double scanning lines and the maximum angle of the k-neighborhood are discussed. And the bidirectional projection algorithm is adapted to determine the central axis. Secondly, an adaptive cross-section extraction algorithm based on the local point density is presented and the cubic B-spline curve is selected to fit the cross-section points after a comprehensive analysis of curve fitting algorithms. Finally, the radial and diametric divergence are used to analyze the local deformation position and overall deformation trend. The proposed method is tested on the large-scale storage tank and tunnel point cloud captured by the terrestrial laser scanner. Results show that the proposed method can adaptively extract cross-sections in any position and accurately obtain the deformation information, and the deformation analysis accuracy is less than 3 mm.