The Reseach of the Vehicle Tracking and Illegal Trajectory Discriminant Based on Video
Min Xie, Zhenghua Zhang, Jun Gu · International Journal of Information and Computer Science · 2014
The single Meanshift tracing with RGB model always loses its target under complex condition. The Meanshift and the prediction of kalman combined in tracking moving targets could improve it. According to the initial location, Meanshift and the prediction of kalman are combined to confirm the new location being the starting point of next tracking, then the trajectories of vehicles are shown. The trajectory of vehicle can distinguish whether the vehicles break rules and regulations. Experimental results show that improved tracking algorithm under complex condition has a certain degree of improvement, and the trajectory tracking has a smooth effect, which doesn’t take too much time, and still meets the requirement of real-time. Extraction of vehicle trajectory provides the basis of illegal discrimination. The lanes and the light determine whether the vehicle has run a red light or illegal turn. The detection method can accurately discriminate illegal vehicles to make video surveillance more