Multi-Lane Detection and Road Traffic Congestion Classification for Intelligent Transportation System

Intelligent Traffic Systems have been widely used for traffic monitoring on roadway, and it is one of the most practicable tools to provide the instant road traffic information for everyone needs it, especially mobile users that demand instant information of road traffic. When traffic congestion arises, if the vehicles get the traffic information earlier, they can choose recommend alternate routes to avoid the traffic jam. Therefore, we propose a traffic congestion classification framework to allow classification of congestion in traffic video sequences from real-time surveillance. The framework consists of three procedures: the first one is the roadway mask with bidirectional roadway analysis, and then the virtual detectors are set up for each lane without lane marking detection. The second step is to estimate the three traffic parameters: flow, speed and density by virtual detectors. In the last procedure, three traffic parameters are utilized to classify the traffic congestion into four levels accurately. The experiment results show that the framework can perform well on more complicated roadway types and simplified procedure of vehicle tracking to reduce the computational cost caused by complex algorithm. Keywordslane detection; virtual detector; traffic parameter; traffic congestion; intelligent transportation system.