Concealed Dangerous Object Detection Based on a 77GHz Radar

In this paper, we propose a method which can detect dangerous objects such as pistol and knife concealed on pedestrian via a 77GHz radar. The proposed method uses range-FFT and Doppler-FFT to obtain micro-Doppler signature of the concealed object. We extract features from micro-Doppler signature with motion pattern of the pedestrian. A classifier based on Support Vector Machine (SVM) is implemented to distinguish dangerous objects from all detected objects by micro-Doppler features. We set up an experiment platform which uses a PVC dummy to simulate pedestrian to verify our proposed method. Results show that our proposed method can get accuracy of 89.8% and miss alarm rate of 3% among 200 times of detection.