Assessment of Unmanned Aerial Vehicle (UAV) based photogrammetric survey for road alignment using direct geo-referencing technology

Abstract

This study has compared the linear measurements of a road network using Differential Global Positioning System (DGPS)-based survey and UAS-based photogrammetric survey. To achieve this objective, 22 random Ground Control Points (GCPs) have been marked along the road network of 1.6 km length. The coordinates of the GCPs have been identified using DGPS and taken as base measurements. Coordinates of the GCPs have also been obtained through aerial images captured under various conditions, depending on Ground Sample Distance (GSD), overlapping of images, light conditions, shutter speed, and ISO of the sensor. The coordinate errors have been expressed as root mean square error (RMSE) relative to the base coordinates obtained through the DGPS and photogrammetric surveys. The results have demonstrated RMSE values as low as 0.263 mm, 0.305 mm, and 0.399 mm for the x, y, and z axes, respectively, under most test conditions, indicating sub-decimeter level accuracy. The selection of 22 random GCPs has been based on a stratified sampling approach to ensure comprehensive coverage and accuracy. It has been observed that the often neglected camera settings can highly affect the location of the GCPs. Therefore, proper sensor settings must be considered while using UAS-based photogrammetric surveys.