Microwave staring correlated imaging via the combination of nonconvex low-rank and total variation regularization

Abstract

Based on the temporal-spatial stochastic radiation field (TSSRF), microwave staring correlated imaging (MSCI) can achieve high resolution images of the targets. This paper focuses on the nonlocal and local regularization to improve the MSCI reconstruction performance. Low-rank regularization is considered to reveal the global information of the images. For local, total variation is a typical choice for its excellent edge preserving ability and noise reduction. Therefore, a method with the combination of the low-rank and total variation regularization is considered in this paper, in which a logarithmic function is taken as a non-convex approximation of the rank function. By elaborately adjusting the regularization parameters, the whole problem is still convex. Thus the Split Bregman method is utilized to solve the convex optimization problem and form the proposed algorithm, namely the LogRankTV algorithm. The effectiveness of the LogRankTV algorithm is demonstrated via the simulation results.