Accuracy Assessment of Aerial Photographic Data from Unmanned Aerial Vehicles Using Ground Control Point Patterns for Route Mapping

Abstract

This study evaluates the accuracy of aerial photographic data obtained from unmanned aerial vehicles (UAVs) using a total of 45 ground control and checkpoint locations. Ground control points along roadsides and riverbanks vary according to the processing model, whereas eight points on mountainous terrain are consistently designated as checkpoints. The ground sampling distance (GSD) is set at 2.36 cm per pixel, with an overlap and sidelap of 80% and 70%, respectively. Nine processing models were analyzed, categorized by three flight path configurations (three, two, and one flight lines). Each configuration incorporated three ground control point distributions: A, B, and C. Data processing was conducted using Pix4D Mapper to generate 3D point clouds, digital elevation models, and orthorectified images. The results indicate that models 3-A and 3-B provide the highest positional accuracy, suitable for mapping at a scale of 1:100 or smaller, meeting the accuracy criteria set by the Department of Lands. Ground control distributions A and B exhibit similar positional accuracy, while distribution C yields the lowest accuracy. Additionally, the number of flight lines directly correlates with positional accuracy, with three flight lines producing the most comprehensive imagery, followed by one flight line, and two flight lines providing the least complete image coverage.