Accuracy Assessment of Unmanned Aerial Vehicle (UAV) Structure from Motion Photogrammetry Compared with Total Station for a Deformed Slope

Abstract

Unmanned Aerial Vehicle (UAV) has been widely used for slope stability analysis. The objective of this research is to test the digital surface model (DSM) results generated from UAV images with the data acquired from total station for a deformed slope. A slope along the Pan Borneo Highway was selected for the study. The UAV survey was undertaken by utilizing DJI Inspire 1 with Zenmuse X3 Gimbal. A total of 10 ground control points (GCPs) were marked during the surveying for validation purposes. Structure from motion (SfM) technique adopted Pix4D enterprise version 4.3.33 to stitch the images for the production of orthophotos and DSM. The root mean square error (RMSE) of the GCPs were checked, where the horizontal RMSE in x direction and y direction are 1.4 cm and 1.8 cm respectively while RMSE in z direction is 2.6 cm. The total station surveying was taken at various locations of slope, which include slope surface with slight to moderate deformation, slope surface with severe deformation, surface channel and at the edge of surface channel. The elevations of DSM results were tested with those surveying data acquired from site. The results show that for slope with slight to moderate deformation, the accuracy of the RMSE in elevation of 4.2 cm can be achieved. Similar RSME accuracy can be attained for surface channel which is 5 cm. However, the RMSE for slope portion with severe deformation is 10.6 cm. From this research, it is found that the UAV-based DSM lower accuracy will be attained for locations of sharp changes in elevation.