A Vision-Based Road Surveillance System Using Improved Background Subtraction and Region Growing Approach

Monitoring traffic intersections in real time is an important part towards Intelligent transportation system. A road surveillance system using background subtraction and threshold segmentation calculated traffic parameters. First an improved background establish and updating algorithm was applied to setup a robust background filtering illumination disturbance. Then a self-adaptive max variance threshold algorithm was adopted to determine the threshold. Third, a fast region growing algorithm was employed to border vehicles in some rectangles fixed their edges. Finally, algorithms were mentioned to calculate traffic parameters including traffic flux, average speed and duty ratio of road. The experimental results show that the system is adapted to monitor a multi-lane road near an urban intersection.