Traffic Flow Detection Using Thermopile Array Sensor

The Intelligent Transportation System (ITS) has been developed rapidly in recent years. As a significant part of the ITS, the collection of traffic information relies on the detection of vehicles by using various sensors, in which micro-electro-mechanical system (MEMS) sensors have been widely investigated. A representative of MEMS sensors is magnetoresistance sensor with stable performance. However, it is hard to detect very low-speed congested traffic. The MEMS thermopile array sensor currently used for human detection can detect stationary infrared objects, including the ignited vehicles. It could be an alternative to detect the vehicles under heavily congested traffic condition. This paper proposes a traffic flow detection method using such a thermopile sensor by the roadside. Traffic flow data are collected and analyzed through a procedure of data preprocessing, background removal, morphology operation, and vehicle detection. A visual background extractor (ViBE) is used in background removal and a finite-state machine (FSM) is designed to reduce the detection error. Different experiments test the detection method in the absence and presence of high temperature interference. The experimental results indicate that the proposed method can successfully detect the traffic flow and it is suitable for occasions where the ambient temperature differs obviously from the vehicle temperature.