Semi-automatic Identification of Rock Discontinuity Orientation Based on 3D Point Clouds and Its Engineering Application

The rock discontinuity orientation has a significant impact on the engineering stability. This paper proposes a semi-automatic method for identifying the orientation of rock discontinuities based on 3D point clouds. First, the normal of each point is calculated by k-nearest neighbors (KNN) and least squares (LS) algorithms. Then the point normal is regarded as the input parameter and the number of joint sets can be determined by density-based spatial clustering of applications with noise (DBSCAN). For each joint set, the point coordinate is considered as the input parameter, and DBSCAN is employed again to obtain the individual joint plane. Finally, the orientation of each joint plane is computed by LS algorithm, and the weighted mean of the orientation based on size is taken as the orientation of each joint set. The correctness of the method is verified by two standard geometric solid objects, and it is applied in real slope engineering to identify the orientation of rock discontinuities. The results are consistent with those observed in the field, and the differences with other methods are discussed. In comparison with the open source procedure DSE and the commercial software ShapeMetrix, the proposed method shows remarkable advantages in terms of precision and accuracy, and can preserve more point cloud information. This study extends the method of semi-automatic identification of rock discontinuities and can provide a reference for similar engineering.