Vision Navigation with Salient Features for Autonomous Aircraft

A novel technique of vision-aided navigation for autonomous aircraft is presented in this paper. The aircraft’s position and pose are estimated with several control points. The saliency descriptor of corner is defined and the control points are selected according their saliency. Control points are tracked in sequential images based on Fourier-Melline transform. The unscented Kalman filter is used to fuse the aircraft state information provided by the vision system and the inertial navigation system. Experiments show that the accuracy, efficiency and robustness of aircraft navigation system are improved with the proposed method.