Ground-penetrating radars (GPRs) are ultra-wideband microwave sensors mainly used for detection and identification of mines and other explosive objects buried in soil or hidden under road construction layers. In this study, we investigated effects of having different construction layers over the soil surface in buried object detection problem using A-scan GPR data. A novel preprocessing technique that makes use of cumulative energy curves of A-scan signals is used for preprocessing and target detection in one easy step. Once we detect a buried object, the next step is to classify this object as either threat or clutter. Target feature extraction is essential for target classification. Simulated GPR data are analyzed in this work by quadratic time-frequency transformation techniques to obtain target features based on electromagnetic signal power distribution in joint time-frequency domain.
展开▼