AI-Driven Method for Wideband GPR Microwave Imaging of Buried Scatterers

An innovative Artificial Intelligence (AI)-driven methodology for the microwave imaging (MI) of buried scenarios with Ground Penetrating Radar (G P R) data is presented. Leveraging the System-by-Design (S b D) paradigm, the complexity of the arising inverse scattering problem (ISP) is addressed by means of a divide-et-impera approach in which a set of interconnected 'functional blocks' are responsible to efficiently tackle specific sub-tasks of the original (fullwave/quantitative) problem. To this end, a smart definition of the GPR-ISP as well as an AI-driven qualitative pre-estimation and a machine learning (M L)-enhanced space exploration techniques are exploited to effectively address the nonlinear GPR-ISP in a competitive way. An illustrative result is shown to assess the effectiveness of the proposed method as well as validate its performance under different scenarios.