An image reconstruction method using GPR data

The spatial resolution of a ground penetrating radar (GPR) system using tra ditional interpretation methods is largely dependent on the isolations of e choes. It is difficult to identify weak echoes that are mixed with either d irect waves or other larger echoes; especially, when the media is absorptiv e, the reflections cannot be identified from the tail of the direct wave. I n this paper, we present an iterative algorithm to reconstruct the image of the subsurface using time-domain waveforms, The forward modeling employed in the algorithm is derived from the transmission line matrix (TLM) method with a perfect matched layer (PML) absorbing boundary. One-dimensional (1-D ) inversion algorithm is briefly described, based on which, a two-dimension al (2-D) inversion method is developed. The inversion algorithm proceeds it eratively in space and time step by step. The inversion algorithm is tested in cases in which both 1-D and 2-D variations exist. The developed inversi on method is used to reconstruct the images of buried objects using GPR dat a with echoes buried in the direct waves. The results show a superior resol ution and stability.