Anonymous vehicle identification on seismic spectrograms Available to Purchase

Real-time monitoring of human movement in urban area is vital for traffic management and assessment of usage of roads. Traditional monitoring methods based on cameras or private smartphone raise severe privacy concerns. With its natural attribute of anonymity, the wavefield recorded by seismic sensors in cities has been shown to have a high coherence with traffic flow. To use seismic data for traffic monitoring, it is important to achieve interpretive traffic information rapidly and accurately from seismic characteristics. We analyze the seismic signals recorded by a group of wireless geophones deployed on the campus of National University of Singapore and by a highway. We observe strong energy across a wide frequency band (5-200 Hz) on the seismic spectrograms from road vehicles, based on which we propose an automatic motor traffic identification algorithm using the high frequencies (30-150 Hz). Cross validated with traffic video, the proposed method shows high accuracy to identify buses, trucks, cars and motorbikes. Our study suggests that the anonymity of seismic method enables high resolution monitoring of traffic with high frequency seismic sensors.