An efficient adaptive approach to automatically identify rock discontinuity parameters using 3D point cloud model from outcrops

Abstract

Digital geological survey methods have become supplementary approaches for traditional geological survey in the last two decades. In this paper, the Unmanned Aerial Vehicle (UAV)‐based photogrammetry technology is used to obtain the 3D point cloud model of rock outcrops. The clustering algorithm is used to automatically identify the rock discontinuity parameters. However, the obtained 3D point cloud model with high resolution often has a huge point data which usually poses a great challenge to the computational efficiency of the automatic identification. In fact, too‐dense point cloud data may not be necessary for cases when the rock mass is relatively intact. The optimal point cloud resolution, which balances the accuracy and efficiency, depends on the degrees of fragmentation of the rock mass under investigation. For a model with the same resolution, large‐size discontinuities may have quite a few redundant point cloud data that are of little use to improve the identification accuracy whereas small‐size discontinuities may not be properly identified due to insufficient number of data points. In this paper, the influence of the degree of fragmentation of rock mass on the identified results was investigated. The uniform grid method was adopted to sparse the 3D point cloud model. The optimal point cloud resolution for different discontinuities was suggested. The applicability and feasibility of the proposed approach were verified via three illustrative examples of typical rock slopes.