Clutter removal in the automatic detection of concealed weapons with late time responses

It has been shown that the late time response of a weapon contains important information about its resonant signature, which is dependent on the weapon's size, shape and constitutive parameters. We extract the resonances of a radar return using the Total Least Squares Matrix Pencil Method. Background clutter is suppressed by de-embedding its eigenvalues from the radar return. An Artificial Neural Network Classifier is used to determine if a target is a threat or not. Finally, we demonstrate the algorithm performance using measured data taken in a cluttered environment. The experiments show that, with proper clutter removal, monostatic radar measurements from 0.5 GHz to 5 GHz within 1.5 meters of the inspected person provide a feasible tool for concealed weapon detection. © 2013 EMA.