Automatic identification of geometrical features representing rock discontinuities from 3D point clouds

Characterizing rock mass discontinuities is a crucial aspect of rock engineering. Current automatic methods for discontinuity identification from 3D point clouds generally do not provide mathematical descriptions of individual 'joints'. To overcome this limitation, this article proposes a three-stage approach with the following steps: (i) identification of 'discontinuity sets' or `families'; (ii) identification of 'clusters', defined as continuous sets of nearby points that belong to a given discontinuity; and (iii) identification of `individual joints,' defined as one or more 'clusters' whose spatial position suggests that they conform to a single discontinuity, even without fulfilling a continuity condition. The process concludes with a mathematical representation of joints using convex hulls, which can be employed to characterize the rock mass. The proposed approach is validated using synthetic and real cases. Results illustrate that the developed tools can efficiently identify convex hulls that represent individual rock discontinuities from 3D point clouds.