Cluster Adaptive Matching Pursuit for Multipolarization Through-Wall Radar Imaging

Abstract

Among various types of sensors, through-wall radar is a promising sensor that can penetrate obstacles and sense the targets behind them. In this article, by exploiting the common sparsity property of the multipolarization images, we propose a cluster adaptive matching pursuit (CAMP) algorithm for multipolarization through-wall radar imaging (TWRI). Our proposed CAMP algorithm is developed from compressive sensing (CS) theory. In the proposed CAMP algorithm, the common sparsity among the multipolarization images is utilized through a multiple-measurement vector (MMV) model. The target area is obtained through a support set. We model the support set as a combination of several clusters, thus the number of clusters and the pattern of each cluster are two important factors that can describe the support set. To obtain the pattern of a cluster that corresponds to a certain target in the imaging scene, we propose an adaptive clustering strategy based on the backtracking scheme. On this basis, the number of clusters is adaptively selected through a two-layer matching pursuit method. Different from conventional matching pursuit algorithms that often use atom methods, the support set is acquired cluster by cluster in the two-layer matching pursuit method, which accelerates the speed of acquiring the support set. Experimental results show that our proposed algorithm can locate targets accurately in TWRI with a high target-to-clutter ratio (TCR) performance.