Reconstruction of shield tunnel lining using point cloud

This paper proposes a technique for automatically identifying segments and creating parametric as-built building information models (BIMs) of shield tunnel lining using terrestrial laser scanning data. The developed algorithm includes two parts: (i) a robust method for the classification of the point cloud of shield tunnel segments, and (ii) a new cylinder fitting method for the high-density incomplete cylindrical point cloud data. The point cloud classification includes i) the generation of the 2-D unwrapped depth map, ii) the edge detection, iii) the regional growth of the tunnel, and iv) the classification. The cylinder fitting consists of i) fitting with known cylinder direction, ii) obtaining the direction of the cylinder axis, and iii) finding the boundaries of the tunnel segment. The proposed algorithm is utilized to analyze two shield tunnels; the results confirm that the developed algorithm is practical, and the root-mean-square error of the point-to-model distance is only 1 mm. •A framework is developed to reconstruct as-built BIM for shield tunnels.•A method is devised to classify the tunnel segments on a 2-D unwrapped depth map.•A new cylinder fitting algorithm for cylindrical point cloud data is proposed.•Parameters for creating segment model are extracted from point cloud.