Study Of Image Aided Navigation For UAV Landing

Runway detection is a key technique during the UAV landing image navigation, which contributes greatly to promoting the development of navigation technique and reducing accidents during landing.In this thesis, in order to position the runway precisely during the UAV landing aiming at ensuring landing safety, the method of airport runway detection and tracking is researched with the help of the landing images which are achieved in the process of the UAV landing, The method includes the algorithm of runway detection and tracking on the foundation of monocular vision. Finally, the validity of the algorithm is confirmed by obtaining the flight parameters.For the runway images in the daytime, a method for automatic recognizing airport runway based on the knowledge of vision is presented. This method makes full use of the characteristics of the runway region and boundaries. The time of Hough transform is reduced, and the request of real time is ensured; then choosing suitable threshold to delete the disturbance point, and the request of veracity is ensured. The test results prove that this method can recognize the runway quickly and accurately.For the runway images at night, the iterative threshold is used to segment image. Therefore, characteristic points of runway light are extracted. Furthermore, when the area of runway is small, by detecting the three points of intersection of runway boundary line, the runway region is confirmed. While the area of runway is big, the runway is recognized by the method of searching characteristic light of runway boundary.Finally, according to the geometry relations between runway boundary and horizon, attitude angle and altitude estimation algorithms based on runway boundaries and horizon are applied to acquire the value of attitude angle and altitude. Moreover, the results from the runway images achieved by camera demonstrate that this method is effective.