Enhanced through-the-wall radar imaging using Bayesian compressive sensing

In this paper, a distributed compressive sensing (CS) model is proposed to recover missing data samples along the temporal frequency domain for through-the-wall radar imaging (TWRI). Existing CS-based approaches recover the signal from each antenna independently, without considering the correlations among measurements. The proposed approach, on the other hand, exploits the structure or correlation in the signals received across the array aperture by using a hierarchical Bayesian model to learn a shared prior for the joint reconstruction of the high-resolution radar profiles. A backprojection method is then applied to form the radar image. Experimental results on real TWRI data show that the proposed approach produces better radar images using fewer measurements compared to existing CS-based TWRI methods.