Reconstruction of extreme topography from UAV structure from motion photogrammetry

The development of unmanned aerial vehicle photogrammetry over the last decade has allowed terrain that is very difficult for humans to access to be captured at very high spatial and temporal resolutions. This paper deals with the application of this technique to the study of extreme topography in a near-vertical road cut-slope. Three photogrammetric projects were carried out: one derived from images taken with the camera oriented horizontally, one derived from images taken with the camera tilted at 45 degrees, and one derived from both sets of images. Point clouds and orthophotos were generated for each of these projects. The best accuracies were achieved by the photogrammetric products derived from the combined images set, which had RMSE equal to 0.053 m, 0.070m and 0.061m in X, Y and Z direction, respectively. A software program was developed to generate contour lines and cross-sections derived from the point cloud, which was able to represent all terrain geometric characteristics, such as several Z coordinates for a given planimetric (X, Y) point. Furthermore, comparing the contour lines and cross-sections generated from the point cloud using the program developed in this project to those generated from the digital surface model showed that the former are capable of representing geometric terrain characteristics that the latter cannot.