Hybrid matching pursuit for distributed through-wall radar imaging

Abstract

In this paper, we consider the problem of through-wall radar imaging (TWRI) with an antenna array and develop a distributed greedy algorithm named hybrid matching pursuit (HMP). By dividing all the antenna phase centers into some groups, the task of TWRI can be formulated as a problem of jointly sparse signal recovery based on distributed data subsets. In TWRI applications, existing distributed greedy algorithms such as the simultaneous orthogonal matching pursuit (SOMP) algorithm and the simultaneous subspace pursuit (SSP) algorithm suffer from high artifacts and low-resolution, respectively. The proposed HMP algorithm aims to combine the strengths of SOMP and SSP (i.e., the orthogonality among selected basis-signals and the backtracking strategy for basis-signal reevaluation) and, accordingly, to reconstruct high-resolution radar images with no visible artifacts. Experimental results on real through-wall radar data show that, compared to existing greedy methods, the proposed HMP algorithm significantly improves the radar image quality, at the cost of increased computational complexity.