Multi-scaling deterministic imaging for GPR survey

The retrieval of the dielectric characteristics of 2-D objects buried in a lossy half-space from Ground Penetrating Radar (GPR) measurements is dealt with. The proposed imaging algorithm is able to effectively mitigate the effects of both non-linearity and ill-posedness of the inversion problem through a multi-scaling deterministic approach based on a Conjugate Gradient minimization (IMSA-CG), while its integration within a Frequency Hopping (FH) strategy guarantees a profitable exploitation of the multi-frequency nature of realistic GPR measurements. The effectiveness of the proposed technique is shown by numerical simulations, and a direct comparison with a single-scaling "bare" implementation is given.