Vision-Based Target Three-Dimensional Geolocation Using Unmanned Aerial Vehicles

This paper develops a method of calculating the three-dimensional (3-D) coordinates of a target with image information taken from an unmanned aerial vehicle (UAV). Unlike the usual approaches to target geolocation that rely heavily on georeferenced terrain database or accurate attitude sensors, the contribution of the proposed method is to offset the constraints and also perform 3-D geolocation of the target based on the relative altitude between the aircraft and the target calculated using the stereo vision technique. Considering the poor performance of the yaw-angle measurement provided by the low-quality attitude sensors in the UAV, a novel vision-based 3-D geolocation method is designed. The proposed method is proven to be valid and practical via simulation results and actual flight experiments.