Nonparametric Bayesian Attributed Scattering Center Extraction for Synthetic Aperture Radar Targets

摘要

As a good way to represent target backscatter measured by high-frequency synthetic aperture radar (SAR) systems, the attributed scattering center (ASC) model is able to provide concise and physically relevant features of a complex target and has played an important role in model-based automatic target recognition (ATR). However, most existing ASC feature extraction methods suffer from imprecise image segmentation or high computational cost, which greatly encumber their practical applications. To tackle this problem, we present a novel ASC feature extraction algorithm for SAR targets based on Lévy random fields in a nonparametric Bayesian framework. Specifically, Lévy random fields, yielding a natural sparse representation of the unknown ASC model, are introduced to construct prior distributions, which lead to the specification of a joint prior distribution for the number of ASCs and the ASC associated parameters. Meanwhile, the problem may be formulated as a sparse representation problem, with regularization induced through the Lévy random field prior. We also develop a reversible jump Markov chain Monte Carlo (RJ-MCMC) method to enable relatively fast posterior inference. Experimental results confirm the effectiveness and efficiency of the proposed algorithm.