Correcting Digital Elevation Models (DEM) from Unmanned Aerial Vehicles (UAV): A New Method Using Polynomial Model Matching Techniques

Abstract

Lee, H.-S. and Oh, J.-H., 2021. Correcting Digital Elevation Models (DEM) from Unmanned Aerial Vehicles (UAV): A new method using polynomial model matching techniques. In: Lee, J. L.; Suh, K.-S.; Lee, B.; Shin, S., and Lee, J. (eds.), Crisis and Integrated Management for Coastal and Marine Safety. Journal of Coastal Research, Special Issue No. 114, pp. 434–438. Coconut Creek (Florida), ISSN 0749-0208. Korean coastal regions have large mudflats that shelter endangered migratory birds and many prosperous ecosystems inhabited by various species living in clusters. To recognize the topographical and environmental characteristics of the coastal regions and to systematically manage them, it is essential to obtain topographical maps of the geographical and environmental characteristics. In particular, the height information of the mudflats is critically important in monitoring changes caused by either natural or artificial influences. Recently, Unmanned Aerial Vehicles (UAV) are shown to be potential candidates in preparing accurate Digital Elevation Models (DEMs) of a terrain. But a major problem in using UAVs is that it is almost impossible to conduct a field survey of Ground Control Points (GCPs). This is because the mudflat areas are not easy to approach places where the proportion of silt and clay content is very high. Without GCPs, the DEM produced from the UAV images and photogrammetric software tend to show a nonlinear distortion due to the aerial triangulation errors. Therefore, we propose a least-square height-difference DEM matching with a polynomial model. This method uses a reference DEM obtained from aerial images to compensate for the location error and the aforementioned nonlinear distortion. Experimental results showed that this significant DEM error modeling was possible with curvilinear translation and constant rotation parameters along the diagonal direction of the test DEM that significantly reduce the location errors and almost eliminate the nonlinear distortion of the UAV-derived DEM.