Automatic extraction of feature points on unmanned airship image

摘要

Unmanned airship of low altitude, being a indispensable supplement of satellite and aerial photogrammetry, it can fly under cloud without airport. It plays a important role in many fields, e.g. disasters monitoring, surveying of large scale and fast change detection. The height displacement of obtained image is big, because of low flying height (only 200 to 600 meters)and topographic inequality, especially the impact of big building of city. If only use gray information of pixel, it cannot meets requirement of image processing, because of small number of tie points between images. In fact, people tend to use feature extraction. Thus, how to obtain sufficient feature fast and efficiently, it brings direct impacts on image matching and application. A mixed algorithm on extraction of feature points is proposed in this paper, which use a improved Forstner algorithm and Pyramid data structure.