Intelligent Identification and Extraction of Geometric Properties of Rock Discontinuities Based on the Multi-View Stereo Technology: A Case Study of the Bimoyuan Tunnel, Sichuan Province, China
摘要
Exploring an effective method that can accurately measure and estimate the discontinuities of the excavation section is crucial for tunnel stability evaluation. In this paper, we propose an intelligent method to extract tunnel face traces: using MVS + SfM to reconstruct the 3D scene of tunnel face, combining KNN algorithm, KDE algorithm and DBSCAN algorithm to realize discontinuous tunnel face point cloud clustering, and using convex hull algorithm to extract tunnel face traces. The original tunnel face image and the trace map are used as training data and label data respectively to train DeepSegNet network with automatic tunnel face trace extraction capability. The method is applied to the Bimoyuan tunnel section of Leshan-Xichang Expressway, and the validation results show that the method can extract the tunnel face traces well, which provides a new idea for the intelligent identification of discontinuities in tunnel face.