A compressive sensing-based computational method for the inversion of wide-band ground penetrating radar data

This work presents an innovative computational approach for the inversion of wideband ground penetrating radar ( GPR ) data. The retrieval of the dielectric characteristics of sparse scatterers buried in a lossy soil is performed by combining a multi-task Bayesian compressive sensing ( MT-BCS ) solver and a frequency hopping ( FH ) strategy. The developed methodology is able to benefit from the regularization capabilities of the MT-BCS as well as to exploit the multi-chromatic informative content of GPR measurements. A set of numerical results is reported in order to assess the effectiveness of the proposed GPR inverse scattering technique, as well as to compare it to a simpler single-task implementation. (paper)