Recent advances in the use of remote sensing techniques allow the acquisition of dense 3D information helpful for the characterization of the rock mass joints. This implies the necessity of having robust and reliable methods to evaluate and extract the primitive geometries representing discontinuities on a rock outcrop. Moreover, these methods have to be easy to use, fast and accurate, which leads nowadays to the tendency of developing automated methods, often having limitations as concerns processing time, definition of parameters and, especially, accuracy. We present here an alternative approach based on two new semi-automatic algorithms, the Iterative Pole Density Estimation (IPDE) and the Supervised Set Extraction (SSE), used in combination with well-known and suitable clustering and density estimation methods. The IPDE performs an analysis based on a threshold value, within which it searches for coplanar points in a range of tolerance, automatically eliminating those below the established threshold, and then finding principal orientations by Kernel Density Estimation (KDE) and identifying clusters by a manual evaluation or through automated clustering methods. The SSE is a tool that allows to extract discontinuity sets from point clouds through an approach aimed to combine observations made in situ with digital results, taking into account the crucial importance of traditional analysis by an expert user. The method was tested in Campania (Italy) at the Cocceio Cave and at the Cetara Tower cliff: at the cave, we were able to recognize an additional set, not identified during previous digital analysis. In the second case, a fully automatic technique, with little or no human intervention on the point cloud, was compared with a previously made supervised method to perform the semi-automatic approach, eventually checking both results with those from traditional surveys, which led the whole analysis to shift the focus on the combined method proposed.
Alternative methods for semi-automatic clusterization and extraction of discontinuity sets from 3D point clouds
Cardia S.
;Parise M.
2023-01-01
Abstract
Recent advances in the use of remote sensing techniques allow the acquisition of dense 3D information helpful for the characterization of the rock mass joints. This implies the necessity of having robust and reliable methods to evaluate and extract the primitive geometries representing discontinuities on a rock outcrop. Moreover, these methods have to be easy to use, fast and accurate, which leads nowadays to the tendency of developing automated methods, often having limitations as concerns processing time, definition of parameters and, especially, accuracy. We present here an alternative approach based on two new semi-automatic algorithms, the Iterative Pole Density Estimation (IPDE) and the Supervised Set Extraction (SSE), used in combination with well-known and suitable clustering and density estimation methods. The IPDE performs an analysis based on a threshold value, within which it searches for coplanar points in a range of tolerance, automatically eliminating those below the established threshold, and then finding principal orientations by Kernel Density Estimation (KDE) and identifying clusters by a manual evaluation or through automated clustering methods. The SSE is a tool that allows to extract discontinuity sets from point clouds through an approach aimed to combine observations made in situ with digital results, taking into account the crucial importance of traditional analysis by an expert user. The method was tested in Campania (Italy) at the Cocceio Cave and at the Cetara Tower cliff: at the cave, we were able to recognize an additional set, not identified during previous digital analysis. In the second case, a fully automatic technique, with little or no human intervention on the point cloud, was compared with a previously made supervised method to perform the semi-automatic approach, eventually checking both results with those from traditional surveys, which led the whole analysis to shift the focus on the combined method proposed.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
