Recognition By Matching With Edge Location And Orientation

This paper describes techniques to perform efficient and accurate recognition in difficult domains by matching edge pixels with associated local orientations. We use a modified Hausdorff measure to determine which positions of each object model are reported as matches. A search strategy is described that allows these positions to be found efficiently and includes techniques to reduce the search time when multiple objects are considered. We have applied this system to recognition in intensity and infrared images. The results indicate that these techniques allow recognition to be performed in complex images efficiently and with a low suite of false positives. Initial experiments on data with ground truth indicate that the separability of the object from clutter is increased when using this information.