Validation of fracture data recognition in rock masses by automated plane detection in 3D point clouds

• Published in: International journal of rock mechanics and mining sciences (Oxford, England : 1997) Vol. 109; pp. 19 - 31
• Main Authors: Drews, T., Miernik, G., Anders, K., Höfle, B., Profe, J., Emmerich, A., Bechstädt, T.
• Format: Journal Article
• Language: English
• Published: Berlin Elsevier Ltd 01-09-2018; Elsevier BV
• Subjects: 3D point cloud; Algorithms; Computer applications; Computer programs; Data analysis; Divergence; Fracture intensity; Fracture mechanics; Geological outcrop acquisition; LiDAR; Mathematical analysis; Outcrops; Plane detection; Planes; Segmentation; Software; Spherical-orientation statistics; Statistical analysis; Three dimensional models; Validation studies; Virtual reality; Geological outcrop acquisition; 3D point cloud; Spherical-orientation statistics; Plane detection; Fracture intensity; LiDAR
• ISSN: 1365-1609; 1873-4545
• DOI: 10.1016/j.ijrmms.2018.06.023

This paper presents (1) an automated method to extract planes and their spatial orientation directly from 3D point clouds, followed by (2) extensive validation tests accompanied by thorough statistical analysis, and (3) a fracture intensity calculation on automatically segmented planes. For the plane extraction, a region growing segmentation algorithm controlled by several input parameters is applied to a point cloud of a granite outcrop. Within its complex surface shape, more than 1000 compass measurements were conducted for validation. In addition, digitally handpicked planes in the software Virtual Reality Geological Studio (VRGS) were used for single plane comparison. In a second test site, we performed fracture intensity calculation in Petrel based on results of the segmentation algorithm on mechanical layers of a clastic sedimentary succession. The comparison of automated segmentation results and compass measurements of three different plane sets shows a deviation of 0.70–2.00°, while the mean single plane divergence amounts to 4.97°. Hence, this study presents a fast, precise, and highly adaptable automated plane detection method, which is reproducible, transparent, objective, and provides increased accuracy in outcrops with rough and complex surfaces. Moreover, output formats of spatial orientation and location of planes are designed for simple handling in other workflows and software.