Abstract: Discontinuity traces constitute linear features generated at the intersection of rock mass surfaces and rock discontinuity, serving as a fundamental and critical parameter for characterizing the strength of rock masses. Quick and accurate identification of the traces of rock mass discontinuities is of great significance for the evaluation of rock mass stability. This paper proposes a novel method for automatically identifying discontinuity traces from 3D point cloud data. Firstly, the potential feature points of the trace are identified based on the differences in the roughness features of the point clouds. Then, the L1 median algorithm is employed to refine the potential feature points and extract the trace skeleton points. This process eliminates the need for point cloud triangulation and can directly process the raw data. Finally, an improved skeleton line growth algorithm is used to connect the discrete skeleton points to generate a continuous trace line. The effectiveness of the new method is validated via four point cloud datasets, and the recognition results are compared with those obtained using CloudCompare and existing methods. The results show that the new method performs exceptionally well in trace line extraction with a high degree of automation, thereby providing effective data support for rock mass quality assessment and stability analysis.

Key words: discontinuity trace, 3D point cloud, intelligent recognition, rock mass discontinuity