High-Resolution DEM Generation of Railway Tunnel Surface Using Terrestrial Laser Scanning Data for Clearance Inspection

Abstract

The current clearance inspection of a railway tunnel is mostly limited to extracting and analyzing 2D cross sections. A novel clearance inspection technique has been introduced to generate the high-resolution digital elevation model (DEM) of a railway tunnel surface (bare lining) from terrestrial laser scanning (TLS) data. In this process, the tunnel boundary points and the upper boundary points of the inner rail are extracted from tunnel point clouds. By using these two boundary point groups, multimodel fitting is used for the estimation of the horizontal and vertical alignments based on the random sample consensus (RANSAC) algorithm, and the final alignments are determined using a global optimization based on piecewise fitting. The theoretical surface of a curved tunnel is reshaped into a developable surface using a new coordinate system named tunnel stationing coordinate system (TSCS). The stationing and cross-sectional horizontal datum (SCHD) and the overbreaking and underbreaking vertical datum (OUVD) are defined on this developable surface and used for DEM generation. A two-stage algorithm for holes filling and nonlining points removal is used to generate to bare-lining DEM. The proposed technique was applied to the point cloud dataset that was acquired in the Xieziyan railway tunnel in Sichuan, China. In the results, DEMs were generated at the resolution of 0.005 m. Compared with total station surveying and an existing filtering algorithm, the bare-lining DEM generation offers high accuracy and more reliable surface modeling for clearance inspection.