Abstract:
Photogrammetry-based methods using everyday, easily available photographic equipment and
free software have gained relevance for roughness measurement of rock joints. However, the
influence of some aspects, such as the number and orientation of camera poses, the parameters of
the Structure-from-Motion (SfM), Multi-View Stereo (MVS) and meshing algorithms, which
influence the resolution and accuracy of the reconstructed models, require proper appraisal. To
assess the surface roughness of a granite rock joint specimen using such photogrammetry
methods, 3D models generated using different settings (i.e., camera sensors, camera poses,
open/free software and workflows), were compared with a reference model obtained by contact
digitization. The results suggest that comparable results can be achieved if the photos adequately
cover the specimen and, at least, equivalent vertex densities are attained. Percentiles p5 and p95
of surface deviations of the tested models to the reference mesh were, in general, lower than
±0.1 mm. Results also showed that this methodology is far more accessible, easier to use, faster
to implement, and less expensive than most currently available equipment and approaches.