A feature-based real-time traffic tracking system using spatial filtering

The detection of traffic parameters has been remaining an important task in ITS. Faced with such a requirement, more and more people resort to vision systems that have been widely installed for monitoring traffic. The paper focuses on the design of a feature-based tracking system, which has an advantage in object segmentation. The method can be briefly described as the following. First, video image data of traffic flow are captured by cameras installed by the roadside; then, features of moving objects are defined and detected; next, during the tracking process, a spatial filter is devised to detect the velocity of the traffic flow. The new position of the object is predicted by using this parameter; finally, these tracked objects are grouped together for segmentation. Simulation results follow for verification