Automatic Recognition of Tunnel Face Rock Information Using Terrestrial Laser Scanning (TLS)
X-Z. Du and T. Qiu
Three-dimensional (3-D) terrestrial laser scanning (TLS) can acquire tunnel point cloud data which contains a wealth of engineering geological information. A method is proposed herein to extract tunnel information from point clouds, including tunnel face extraction, cosine clustering, local multi-layer region recognition and correction algorithms. Specifically, the extraction algorithm is used to extract the tunnel face from the complete tunnel point clouds. The normal is then estimated through the adjacent points of the point clouds and the preliminary cosine clustering algorithm is performed on the estimated surface normals. In the local multi-layer area, the correction algorithm divides the point clouds plane into three layers and corrects the wrong point normal direction based on the global semantics. The results show that the accuracy and reliability of the algorithm are superior to the previous methods.
Keywords: Terrestrial laser scanning (TLS), tunnel, point cloud, normal orientation, automatic recognition, clustering