Clutter reduction using background subtraction of Ground Penetrating Radar for landmine detection

GPR (Ground Penetrating Radar) is an application of electromagnetics that has been widely gaining momentum for the detection of landmines. Existing transmitter-receiver systems suffer from high background clutter, high false alarm rate and strong reflections from the air soil interface. Although many approaches have been developed, implementing them in man portable system is not practical. Our paper aims at developing and testing a viable signal processing methodology to address these issues. The signal processing technique considered is background subtraction in spatial domain. The major advantage of the approach is treating the raw data as images instead of s-parameter matrices. Three kinds of filters are used, namely averaging, thresholding and localization filter. The algorithms for filter implementations are simple and can be easily incorporated in DSPs. The B-scan data required for testing was simulated by considering frequency in the range 1-1.5GHz with soil permittivity. Considering various cases simulations are carried out using MatLab to reduce clutter using cascaded filter as a background subtraction approach.