Time Reversal Imaging of Deeply Buried Targets Under Moderately Rough Surfaces Using Approximate Transmitted Fields

Abstract

In this paper, for time-reversal (TR) imaging of buried targets under moderate rough ground surfaces, the asymptotic half-space Green's function (AGF) is utilized. For the forward problem, a parallel finite-difference time-domain technique is used, in which the excitation is an array of rectangular apertures and scattered fields are probed by an array very near the ground surface. Using the scattering matrix of the TR operator at a frequency range of 0.5–2.5 GHz, multiple signal classification (MUSIC) images are calculated. Assuming both dielectric and metallic targets and the Gaussian random rough surfaces, TR images using the conventional Green's function of a homogeneous medium with an averaged permittivity of free space and the ground are compared with those using AGF. It is shown that two different targets, at the same range but different depths, can be successfully detected once the AGF is used. Furthermore, using five receiving arrays at different heights from the ground, a good depth resolution is achieved. In this regard, transverse and depth resolutions are studied for ten random realizations of rough surfaces and two different root-mean-square heights (5 and 10 mm), leading to the mean depth resolution of about 0.9 m.