Discrimination of buried objects using GPR responses

This paper describes experimental result of discrimination of buried objects using ground penetrating radar (GPR) data. As a target object, an anti-personnel landmine is considered. Two signal processing techniques are applied to extract a target response from raw GPR data. As a measure for target discrimination, the cross-correlation between waveforms of the extracted target response and a target template that is prepared by prior measurement is employed. Discrimination performance is checked by using a plastic landmine model and six confusing clutter objects that have approximately the same size as the landmine model. The result shows that the cross-correlation is one of the effective measures for this purpose. It, however, also indicates that using only the cross-correlation is not sufficient for accurate discrimination and thus other types of measures (features) for target identification should be added.