Research On Red Light Violation Detection Based On Video-Image

With the full speed development of economy, the application research of Intelligence Transportation System (ITS) comes into being because of increasingly serious traffic problems which produced by the rapid increase of vehicles'quantity. Currently, ITS is the preceding research topic in transportation region applied in electronic information techniques, and our country has arranged it into the emphases region of high tech industry developing in priority.The statistics of many countries reveal that most serious traffic accidents occur at the parting of the ways, and most of these are brought about by red light violation which has became into one of main threatens of road safe. Therefore, people often utilize the methods of burying influence loop sensors cooperated with grasp cameras in the crossing roads. But this method has a good many abuses that it can only be used in fixed crossing detection and it demands to destroy the road when imbedding loops which will influent the normal traffic.The paper presents research on monitor of vehicles peccancy on a image.Two aspects are presents to resolve vehicles peccancy, one is on the recognition of diference, another is recognition of virtual loop. The emphasis includes moving vehicle detection, recognition, tracking etc. In course of moving image detections, a method of the diference between inter frame is proposed to detect moving regions. Also a method of the region growth is proposed to achieve connectivity determinant of moving parts. In course of tracking, a method of vicinity is utilized together with Kalman filter to present moving tracking and velocity parameters of vehicles. Finally,it combines the method of VLD with the method of moving detection to estimate vehicles peccancy. The method is processed simulation experiments. The results indicate if integrating image disposal and recognition in this paper, it can nicely recognise vehicles red light violation in image frequency.