Search

To recreate high-quality three-dimensional (3D) models of rock joints for quantitative roughness characterization more economically, portably, and quickly, it is necessary to assess the feasibility and accuracy of 3D reconstruction of rock joints using SfM & MVS photogrammetry with a camera in animation mode. A low-cost approach for 3D reconstruction by shooting videos was conducted to obtain point clouds of rock joints. Besides, to evaluate the accuracy of photogrammetry, high-resolution models of the same cases were collected using 3D laser scanning technology to compare the difference between models reconstructed by these two technologies, and the difference was calculated by one algorithm—Iterative Closest Point (ICP). Comparative analyses demonstrated that the models reconstructed by photogrammetry and the output of the laser scanning were in good agreement. Furthermore, these two different sources of point cloud data were used to realize quantitative characterization of the structural surface roughness via calculating JRC and θ*max/(C+1) values, and the related assessment results also showed good agreement. To ensure the precision of 3D models of rock joints, it is strongly advised that the 3D reconstruction be carried out using a camera that has been properly calibrated. Moreover, there is a significant difference in calibrated camera parameters between animation mode and image mode, which cannot be ignored in 3D reconstruction. Comparative analyses demonstrated that SfM & MVS photogrammetry with a DSLR camera in animation mode can generate a point cloud of multi-scale rock joints with acceptable accuracies for roughness estimation, proving the effectiveness of a low-cost approach for 3D reconstruction by shooting videos. [ABSTRACT FROM AUTHOR]