Evaluating the accuracy of digital surface models generated from imagery acquired with unmanned aerial vehicles

Abstract

This paper assesses the accuracy of the UAV-based Digital Surface Model (DSM) in a small area with complicated terrain based on the number of ground control points determined by GNSS technology. Experimental results obtained at the mineral exploitation and processing area of ​​Tan Tien Co., Ltd., Yen Bai province show that 5 ground control points are the minimum number needed for image correction to generate DSM with a mean square error of 14.744 cm. In addition, the optimal number of ground control points provides the highest accuracy for DSM models while also meeting the required accuracy for creating a 1:1000 scale map of 20 points. The results of the paper can be used as a reference when choosing ground control points to establish DSM for areas with areas and topographical characteristics similar to the research area of ​​the paper.