Innovative Solutions Based on the EM-Algorithm for Covariance Structure Detection and Classification in Polarimetric SAR Images

Abstract

This article addresses the challenge of identifying the polarimetric covariance matrix (PCM) structures associated with a polarimetric synthetic aperture radar (SAR) image. Interestingly, such information can be used, for instance, to improve the scene interpretation or to enhance the performance of (possibly PCM-based) segmentation algorithms as well as other kinds of methods. To this end, a general framework to solve a multiple hypothesis test is introduced with the aim to detect and classify contextual spatial variations in polarimetric SAR images. Specifically, under the null hypothesis, only one unknown structure is assumed for data belonging to a two-dimensional spatial sliding window, whereas under each alternative hypothesis, data are partitioned into subsets sharing different PCM structures. The problem of partition estimation is solved by resorting to hidden random variables representative of covariance structure classes and the expectation–maximization algorithm. The effectiveness of the proposed detection strategies is demonstrated on both simulated and real polarimetric SAR data also in comparison with existing classification algorithms.