Novel Feature Preserving Method for Simplifying the Surface Point Cloud of Underground Space Tunnel

With the development of intelligent technologies in underground spaces, such as coal mines, 3-D laser scanning technology has been widely applied for monitoring the deformation of surrounding rocks in these mines. However, the large amount of point cloud data obtained through 3-D laser scanning technology is not conducive to efficient storage and rapid processing. Therefore, this article proposes a novel feature preserving method for simplifying the surface point cloud of an underground space tunnel. First, the fusion criterion is constructed based on curvature, normal vector variation degree, and point-to-tangent plane distance. Based on this, feature points are selected. Then, use the octree and cluster the divided set of nonfeature points. Use the farthest point sampling (FPS) method to complete the simplification work in each cluster. In the experimental part, the Fandisk dataset is used to verify the superiority of the feature point selection method. The Bunny dataset is used to show the superiority of the whole algorithm. Finally, use the real underground tunnel data to show the feasibility and superiority of the algorithm in a specific real-world environment. The publicly WHU-TLS dataset and a coal mine dataset (C-M dataset) are used in this part. The C-M dataset is obtained by our measurements under real conditions in Shenmu County, Shaanxi Province, China. When setting, the simplification rate of the right-side point cloud of the C-M dataset is 30%. Compare the standard deviation of the proposed method, the random sampling method (RSM), the voxel grid method (VGM), and the curvature sampling method (CSM). The standard deviation of the proposed method can be reduced by about 35% at most. It is further demonstrated that the proposed method can better maintain the geometric features while ensuring the simplification rate.