Camera calibration for UAV ground feature localization

This paper describes a method to calibrate the geometric position and orientation of a camera mounted onto the underside of an unmanned aerial vehicle (UAV). The method takes as input multiple images captured in flight of a stationary ground feature for which GPS coordinates are known. The UAV's location and pose during each image capture is assumed known from on-board GPS and IMU devices. The ground feature's pixel coordinates in each image are also known (obtained using image processing or by inspection). Using these inputs, the described method uniquely determines the camera's relative position/orientation with respect to the UAV using a nonlinear root-finding algorithm. The method represents a simple, yet effective, calibration procedure that uses information that is readily obtained from a typical UAV flight. Results from test flight experiments are presented.