UAV photogrammetry offersapowerful and cheap methodology to reconstruct the terrain geomorphology. The purpose of this study is the application of a generic Structure-from-Motion workflow to properly elaborate different se
t of images from multitemporal surveys to perform a surfaceand volume change detection by meaning of a cloud-to-cloud comparison. The complex geomorphology of the site of interest, characterized by niches, asperities, flat rock walls with different orientation, debris deposits andisolated rock blocks, challengesthe accurate reprojection of the image points into a 3D space. The chronological comparison of the point cloud offers a qualitative and quantitative estimation of distances and volume change between sequential models. For the cloud-to-cloud distance computation, a level of accuracy accounting for different sources of uncertainty was estimated(nominal precision of each surface, multitemporal comparison between different surfaces with different level of accuracy, noise due to vegetation and other local couses, ...). Seven data sets were available for this study and they were acquired by two different faculties of the University of Ljubljana, Biotechnical Faculty and Faculty of Civil and Geodetic Engineering. Some of thesurveys and some of the drone flights were performed with different approaches, leading to different accuracy in the final reconstruction of the terrain. The data processing has been performed with the latest versions of Agisoft Metashape and CloudCompare software.
|
