Aircraft Recognition in SAR Images Based on Scattering Structure Feature and Template Matching

Abstract

Scattering characteristics of targets are of great importance for synthetic aperture radar (SAR) image analysis. In this paper, a novel template matching based aircraft recognition method with scattering structure feature (SSF) is proposed to improve recognition accuracy and efficiency in SAR images, mainly including the feature model construction stage and the recognition stage. In the former stage, the SSF, being composed of strong scattering point and its corresponding scattering intensity distribution, is extracted by a scattering structure feature model newly defined with Gaussian mixture model. In the recognition stage, template matching is implemented via the proposed sample-decision optimization algorithm. Specifically, in the sample step, a geometric prior based Monte Carlo method with Hausdorff distance is introduced to improve the efficiency of candidate template selection. In the decision step, coordinate translation Kullback–Leibler divergence is proposed by defining a new entropy function of translation coordinates to achieve the goal of translation invariance. Experimental results are given to demonstrate the accuracy and efficiency of the proposed method.