Abstract: The structural plane spacing is an important parameter in rock mass stability and mechanical properties analysis. It is widely used in the numerical computation of rock mechanics， mining engineering， slope monitoring and other fields. Based on the three-dimensional rock mass slope point cloud data acquired by non-contacting surveying method， this paper proposes a structural surface refinement classification method using density clustering. Projection transformation and scattered point fitting are used after a previous step of general discontinuity classification. The refinement classification is used to compute the structural plane spacing and the rock mass joint number. A prototype system for parameter calculation and analysis of structural surface refinement classification and spacing was designed and developed. The real case experiment and analysis show that the proposed method performs effectively in realizing the automatic refinement classification of structural planes， and in calculating the related parameters such as structural surface spacing， which may provide further support for rock mass quality classification and rock mass stability analysis.

Key words: 3D point cloud, structural surface, refinement classification, DBSCAN clustering, structural plane spacing