Automatic extraction of tunnel centerline and cross-sections from 3D point clouds

Abstract

Abstract The accurate extraction of tunnel centerline and tunnel cross-section plays a vital role in many tunnel deformation monitoring projects. To address this issue, An automatic data extraction method for a tunnel centerline and cross-section is proposed based on the three-dimensional laser scanning data. A k-dimensional tree index is first constructed based on the horizontal projection of the tunnel point clouds, and then a series of seed points is selected as the center of several neighborhood point sets with different radii; the difference set region, which will be used to calculate the tunnel horizontal centerline point based on the convex hull operation, is obtained. Then, according to the horizontal centerline, the cross-section of the tunnel point clouds is extracted and fitted to obtain the tunnel cross-section parameters and three-dimensional centerline. Based on the design values of the three-dimensional centerline and tunnel radius, the overall model of the tunnel can be constructed and applied to the visual analysis of tunnel deformation. The experimental results show that the proposed method can accurately obtain the tunnel centerline and cross-section parameters, with a minimal 0.011 cm error for the extraction of the horizontal centerline. The proposed tunnel model could simulate the undulation of the tunnel surface well and provide reference values for similar cases.