Identification and characterisation of rock mass discontinuity sets using 3D laser scanning

In this research a methodology has been worked out to automate the identification and characterisation of rock mass discontinuity sets using 3D terrestrial laser scan data. With 3D laser scan surveys a dense point cloud is generated that represents the geometry of the scanned rock face in very high detail. Through 3D surface reconstruction techniques the original rock surface can be rebuilt in the form of a very large number of 3D facets. The normal vector data representing these 3D facets can be plotted as poles in a hemispherical projection. In order to avoid the subjective choice of discontinuity sets from such a kernel density distribution, this process has been automated using fuzzy k-means cluster analysis. An approach to the number of discontinuity sets in the normal vector data set is obtained using cluster validity indices. Any apparent outliers that are present in the clusters determined are rejected based on statistics, assuming a circular distribution about the mean orientation. All facets that still contribute to a (delimited) cluster are used to extract the individual 3D 'virtual' rock surfaces and corresponding point clouds within each discontinuity set. A straightforward algorithm has been developed for this. By fitting planes to these point cloud data using Principal Component Analysis, it is possible to determine the intersection distances of the discontinuities with a line of any orientation (discontinuity spacing analysis). With this method, it is shown that (near) planar rock surfaces of discontinuity sets captured by the laser scanner can be identified and extracted as point cloud or facet surface d​a​t​a​.​