Skip to Main Content
In this paper, we consider landmine detection using ultrawideband synthetic aperture radar, where the two main challenges are feature extraction and discriminator design. The space-wavenumber processing is proposed to retrieve the frequency-and aspect-angle-dependent scattering features of suspected objects. In order to reduce the dimensionality of the input feature vector for a discriminator, the sequential forward floating selection method is used to choose efficient features. Based on the obtained feature vector, a fuzzy hypersphere support vector machine is designed to deal with the problem of detecting landmines in an unconstrained environment. The experimental results show that the proposed method can achieve a significant improvement in detection performance for antitank mines.