Automatic Measurement of Control Points for Aerial Image Orientation

Abstract

This paper presents an automatic workflow for measuring natural ground control points (GCPs) in aerial images. The approach is based on the acquisition of multi-scale fisheye vertical terrestrial images over distinguishable areas to automatically generate georeferenced image patches. These are matched with aerial images using correspondence techniques and then used as ground control. The technique was assessed using a bundle block adjustment, indicating an accuracy based on check points of approximately 1/5 of the ground sample distance (GSD) in planimetry and 1 GSD in elevation. Automatic GCP measurements thus improved the final accuracy in comparison with manual measurement.