Feature Extraction and Adaption in Micro-Doppler Image Based Concealed Dangerous Object Detection by Millimeter Wave Sensors

Recently, public security becomes more and more important in daily life. Terrorists or extremists can attack common people by small objects such as pistol or knife in densely populated place. In previous work, we proposed a micro-Doppler image based method to detect dangerous objects concealed on walking through persons ubiquitously. The accuracy of the proposal is affected by walking velocity. In this paper, we enhance the method in the aspect of feature extraction to overcome such issue. The proposed method can extract and adapt the changeable position of effective feature in the micro-Doppler image. Experimental results show that accuracy of the proposed method is 90% when detect the persons with different walking velocity.