Recognition of Occluded Targets Using Stochastic Models

R ecognition of o cclude d obje cts in synthetic ap erture radar (SAR) images is a signi cant problem for automatic target recognition. In this paper, we present a hidden Markov modeling (HMM) based approach for recognizing objects in synthetic ap ertur eradar (SAR) images. We identify the peculiar char acteristics of SAR sensors and using these characteristics we develop featur ebased multiple models for a given SAR image of an object. The models exploiting the relative geometry of fe ature locations or the amplitude of SAR radar return are based on sequentialization of scattering centers extracte d from SAR images. In order to improve performance we inte grate these models syner gistically using their probabilistic estimates for recognition of a particular target at a speci c azimuth. Experimental results are presente d using both synthetic and real SAR images.