Extraction of Planar Patches from Point Clouds to Retrieve Dip and Dip Direction of Rock Discontinuities

Laser scanners and image matching algorithms produce large amounts of 3D data, allowing detailed and accurate descriptions of objects. This is true, though, only if interactive or automated software tools are available, to extract the information relevant to the specific task from raw 3D data sets without too much user interaction. In the last two years our group has been involved in applying new surveying methods to determine the dip and dip directions of discontinuities in rock faces; knowledge of these parameters, that can be computed from an high resolution DSM of the rock, allows for a stability analysis to be carried out. In this paper we describe an automated approach to extract from a dense DSM of a rock face, obtained by photogrammetry or by laser scanning, the discontinuity planes of the rock and compute their parameters. A multi-resolution DSM pyramid is generated from the original DSM; in a hierarchical scheme, the DSM is segmented in planes at each level of the pyramid, using RANSAC with topological constraints. Two clustering steps are also executed, to balance the fragmentation of the planes and the accuracy of the surface approximation. Finally, dip and dip directions are computed from the normal vector to each plane. The proposed method, applied to a test site, gave results equivalent to those of a traditional survey.