Detection and identification in the intelligent traffic video monitoring system for pedestrians and vehicles

On the most of highway, tunnels and bridges, pedestrians are not allowed to access in the current public traffic management. The traditional transportation surveillance system can only monitor the scene but could not alarm automatically for the abnormity, so how to detect the pedestrians and alarm automatically when the people have access to the highway is a great challenge for the intelligent transportation video surveillance system. The paper proposes an algorithm which can solve the problems effectively by the improved Gaussian mixture model and Support vector machine. First of all, the paper introduce an improved Gaussian mixture model which can effectively detect the moving objects and resolve the problems of Gaussian mixture model sensitive to light changes. Then the paper designs some classifiers to recognize the pedestrians and vehicles by the idea of the improved SVM. The experimental results show that the method has a high recognition rate and can also satisfy the real-time intelligent transportation surveillance.