Adaptive Plastic Mine Imaging by Complex-Valued Self-Organizing Map

摘要

Plastic mine detection with a metal detector, which has been widely used as a mine detector, is difficult task. Ground penetrating radar (GPR) is promising technology in plastic anti-personnel mine detection. However, conventional radar system doesn't work well because of the low reflectance of plastics and the influence of ground surface. We propose a novel radar imaging system which processes multi-frequency interferometric data adaptively. In the system, we use interferometric radar to pay attention to both of amplitude and phase to detect plastic anti-personnel mines that have low reflectance. The system processes three-dimensional data (two-dimensional image x frequency) using a Complex-valued Self-Organizing Map(C-SOM) that takes into account the neighboring data relations. We report the principle and results of experiments. We demonstrate a successful visualization of plastic mine buried near the ground surface.