Detection and discrimination of landmines in ground-penetrating radar using an EigenMine and fuzzy-membership-function approach

This paper introduces a system for landmine detection using sensor data generated by a Ground Penetrating Radar (GPR). The GPR produces a three-dimensional array of intensity values, representing a volume below the surface of the ground. First, a constant false alarm rate (CFAR) detector is used to focus attention and identify candidates that resemble mines. Next, translation invariant features are extracted by projecting the magnitude of the Fourier transformation onto the dominant eigenvectors in the training data. The training signatures are then clustered to identify prototypes. Crisp and fuzzy k-nearest neighbor rules are used to distinguish true detections from false alarms.