Landslide movement detection using unmanned aerial vehicle (UAV) photogrammetry

Abstract

Nowadays, Unmanned Aerial Vehicle (UAV) technology is increasingly widespread in monitoring and tracking landslides. This technology can also produce high-resolution digital surface model (DSM), digital elevation model (DEM) and orthophoto which can help researchers in detecting land surface change, especially in small areas. In this study, the research methodology was designed by comparing two observation data with different flight altitudes of 50 and 60 meters. The main purpose of this study was to assess the accuracy of the registration method between Match Bounding-box Centers (MBBC) and Iterative Closet Point (ICP). For the first observation data, the data was retrieved in July 2018 while for the second observation data in November 2018. During the data collection process, DJI Phantom 4 with high-resolution camera was used to take images according to predetermined flight planning. Registration process and georeferencing have been implemented using Pix4D software while point cloud comparison has been implemented using CloudCompare software. The Global Navigation Satellite System (GNSS) observation methodology has been implemented for the preparation of ground control points (GCPs) and check points (CPs). Image quality has been proven by comparing check points with GNSS observation method. Based on the results obtained, the RMSE for both observation data ranges from 0.017 meters to 0.040 meters. The results of the registration method show that ICP registration method is better than MBBC registration method in detecting land movement. In fact, flight altitude also plays an important role in assessing accuracy of point cloud. In overall, UAV platform and ICP method is the best option to detect landslide movement in short time and save cost.