Video detection system for traffic Iight sensor

• Main Authors: Rahmat, Riza Atiq, Jumari, K., Hassan, Azmi, Basri, H.
• Format: Article
• Language: English
• Published: Hemming Group 2001
• Subjects: QA75 Electronic computers. Computer science; T175 Industrial research. Research and development
• Online Access: http://irep.iium.edu.my/37311/1/azmi.pdf; http://irep.iium.edu.my/37311/

The point based inductive loop is widely used in conventional traffic light sensors. The sensor is used either to detect the presence of vehicles or : to measure the gap or headway of the arriving vehicle in the vehicle-actuated system or to count the traffic volume and to determine the queue length in a coordinated adaptive system. In a more sophisticated system, the sensor is also used to detect any traffic incident. However, the rising cost of installing the loops and disruption of traffic flows during installation or maintenance has resulted in the video detection system becoming more attractive. In addition, the cost of equipment for the video detection system has reduced substantially in the past l0 years. This paper describes the utilisation of a video camera and image processing to detect the presence of vehicles, to count the volume of approaching traffic, to measure queue length and to detect traffic incidents at the approach road of a signalised intersection. Neural networks were used to detect the presence of the vehicles, to detect the traffic incident and to measure the queue length by identifying whether the road surface was occupied by vehicles and whether these vehicles were moving or stationary for a specified duration of time. The number of arriving vehicles was counted by observing the fluctuation of the selected pixels values in the middle of the traffic lane. A single camera which was developed in this study is able to capture the above mentioned parameters simultaneously from a multi-lane road approach.