Using street view imagery for 3D survey of rock slope failures

Abstract

Abstract. We discuss here the challenges and limitations on surveying rock slope failures using 3D reconstruction from images acquired from Street View Imagery (SVI) and processed with modern photogrammetric workflows. We show how the back in time function can be used for a 3D reconstruction of two or more image sets from the same site but at different instants of time, allowing for rock slope surveying. Three sites in the French alps were selected: (a) a cliff beside a road where a protective wall collapsed consisting on two images sets (60 and 50 images on each set) captured on a six years timeframe; (b) a large-scale active landslide located on a slope at 250 m from the road, using seven images sets (50 to 80 images per set) from five different time periods with three images sets for one period; (c) a cliff over a tunnel which has collapsed, using three images sets on a six years time-frame. The analysis includes the use of different commercially available Structure for Motion (SfM) programs and comparison between the so-extracted photogrammetric point clouds and a LiDAR derived mesh used as a ground truth. As a result, both landslide deformation together with estimation of fallen volumes were clearly identified in the point clouds. Results are site and software-dependent, as a function of the image set and number of images, with model accuracies ranging between 0.1 and 3.1 m in the best and worst scenario, respectively. Despite some clear limitations and challenges, this manuscript demonstrates that this original approach might allow obtaining preliminary 3D models of an area without on-field images. Furthermore, the pre-failure topography can be obtained for sites where it would not be available otherwise.