Model-based aircraft recognition in perspective aerial imagery

Automatic monitoring of scenes for the purpose of detecting and reporting changes is an important task in mapping and surveillance applications. Recognition of mobile objects in images of aerial scenes in particular, is a difficult task due to the lack of suitable object models and image resolution. For some applications aircraft recognition has to be accomplished in the presence of occlusion, camouflage and various forms of image degradation. This paper presents a recognition technique based on low-level matching between segments in the images and segments of the projection of a three dimensional model of the objects of interest. The model is extracted from one of more views of the scene. Matching segments are analyzed to determine the accuracy of the recognition