Fast GPR underground shape anomaly detection using the Semi-Analytic Mode Matching (SAMM) algorithm

Ground penetrating radar is an effective means of characterizing the subsurface and is an invaluable tool for detecting buried anomalies, including land mines, utility lines, tunnels, and pavement deterioration [1-3]. The conventional strategies of migrating monostatic time domain signals or using synthetic aperture radar (SAR) techniques to produce images are improved in this work by a model-based inversion algorithm which uses the Semi-Analytic Mode Matching (SAMM) method [4,5]. Similar to pattern matching, this approach simulates the scattering from a buried object of specific size and shape and compares it to observed signals. Since scattering is considered from the entirety of the target rather than by analyzing each spatial pixel as in SAR imaging, the inverse-SAMM algorithm is superior at characterizing larger buried objects. Here, we consider a forward-looking, standoff, vehicle-mounted multi-monostatic stepped-frequency radar operating at 0.4-3 GHz searching for buried mines in two dimensions (depth and down-track). Detection and localization of metallic or plastic-cased mines is excellent and size characterization is reasonable, even when fairly substantial ground surface roughness is included.