Moving Human Indication Scheme Using Doppler Radar Sensor

In this paper, we proposed a human classification method based on a Doppler spectrum distribution using CW (Continuous Wave) radar. Because the Doppler spectrum distribution of a moving human is highly variable over time, we proposed three feature: Doppler spectrum extension in the current frame, Doppler spectrum variance between two successive frames, and detected velocity variance feature between two successive frames. To verify the proposed classification method, we obtain the real data using 24GHz CW module consisting of antennas, RF transceiver, and baseband circuit. We conducted two scenarios of walking humans and moving vehicles. From the results of verification, we proved that the proposed method is effective to classify humans based on the three proposed features.