Radar Imaging Through Cinderblock Walls: Achievable Performance by a Model-Corrected Linear Inverse Scattering Approach

We address the problem of imaging targets located behind an inhomogeneous wall made with cinderblocks. The problem, which has relevance in through-wall-imaging applications, is characterized by the presence of multipath propagation phenomena usually producing artifacts and distortions in the retrieved images, if not suitably accounted for in the scattering model. The strategy here adopted to mitigate this issue is to employ a linearized scattering model based on the Born approximation, where the kernel of the relevant integral equation is evaluated numerically by means of the finite-difference time-domain method. In this way, the complexity of the background scenario is accurately taken into account. The inversion is successfully performed by the truncated singular value decomposition algorithm so as to regularize the inverse problem. The achievable imaging capabilities are analyzed in terms of resolution limits, and most notably, resolution can be effectively enhanced, owing to multipath exploitation. Numerical tests based on synthetic data are reported to assess the reconstruction performance in the case of canonical objects.