Adaptive Radar System to Visualize Antipersonnel Plastic Landmines

Abstract

We extend the idea in the adaptive classification described in Chapter 5 by introducing frequency-domain information to visualize plastic landmines buried shallowly underground. Antipersonnel landmines, in particular plastic ones, use so slight metal that it is difficult to detect them with metal detectors because many shots and metal fragments are scattered under battlefields. The shallowness also causes serious surface-reflection noise. We construct a phase-sensitive millimeter-wave / microwave front-end to observe ground reflection in spatial and frequency domains, and feed the data to a complex-valued self-organizing map (CSOM). The CSOM visualizes plastic landmines by segmenting the reflection image adaptively.