Subsurface imaging for anti-personal mine detection by Bayesian super-resolution with a smooth-gap prior

Abstract

Ground-penetrating radars (GPRs) have been studied in order to reconstruct subsurface images. The signal observed by GPR typically includes very strong noise, and reconstruction of the image is a difficult task. We propose a new subsurface imaging method based on the framework of Bayesian super-resolution. In the framework, we can incorporate additional information into the reconstructed image by considering a smooth-gap prior, which can represent the smoothness of the subsurface image and gaps between materials, and which improves the quality of the reconstructed image. We investigated the performance of the proposed method with a synthetic GPR dataset, and confirmed the validity of the proposed method.