Feature matching and target recognition in synthetic aperture radar imagery

An approach for target matching in synthetic aperture radar (SAR) imagery is presented. The method is feature based where feature points in a target candidate are matched against those from an exemplar database. Matching is formulated as a non-linear optimization problem that encourages matches while minimizing the distance between the matched features. The formulation allows for missing, spurious and shifted feature points. A non-linear function is used to convexify the search space to enhance the search for the minimum objective cost. Extensions are presented for the use of two different feature types in the matching. Registration of the images is computed during the matching process in an iterative manner. Matching results are presented for simulated XPATCH and real MSTAR SAR target imagery