Landmine detection using fractional Fourier features

This paper addresses the problem of detecting anti-personnel landmines, in an environment that could contain clutter, using features extracted via the fractional Fourier transform. The focus is on identifying and using fractions of the Fourier transform that lead to better detection performance assuming that the landmine is within a reasonable depth variation below the surface. The performance of the proposed detection scheme is assessed in terms of detection rate and false alarm rate using real landmine signatures. The effects of removing ground clutter are examined, and the effects of knowing the mine depth a priori are also discussed.