An l1-regularized least squares algorithm for reconstructing step-frequency ground penetrating radar images

Abstract

Impulse-based ground penetrating radar (GPR) has been proposed as a way to detect landmines and improvised explosive devices (IEDs). However, a drawback of such radar systems is the difficulty in transmitting a signal with an arbitrary bandwidth and shape. Step-frequency GPR has been recognized as a way to precisely control the bandwidth and spectral shape of the transmitted pulse. In this paper, we extend a previously developed ℓ1-regularized least squares algorithm, which has been successfully applied to impulse-based GPR image reconstruction, to step-frequency GPR. We investigate the performance of the proposed algorithm using simulated step-frequency GPR data. The initial results are promising.