Vehicle Detection and Classification for Low-Speed Congested Traffic With Anisotropic Magnetoresistive Sensor

A vehicle detection and classification system has been developed based on a low-cost triaxial anisotropic magnetoresistive sensor. Considering the characteristics of vehicle magnetic detection signals, especially the signals for low-speed congested traffic in large cities, a novel fixed threshold state machine algorithm based on signal variance is proposed to detect vehicles within a single lane and segment the vehicle signals effectively according to the time information of vehicles entering and leaving the sensor monitoring area. In our experiments, five signal features are extracted, including the signal duration, signal energy, average energy of the signal, ratio of positive and negative energy of x-axis signal, and ratio of positive and negative energy of y-axis signal. Furthermore, the detected vehicles are classified into motorcycles, two-box cars, saloon cars, buses, and Sport Utility Vehicle commercial vehicles based on a classification tree model. The experimental results have shown that the detection accuracy of the proposed algorithm can reach up to 99.05% and the average classification accuracy is 93.66%, which verify the effectiveness of our algorithm for low-speed congested traffic.