Intersection Vehicle Detection System Based On Vision Sensor

With the development of the automobile industry, car penetration continues to increase,traffic congestion has become a major trouble in people’s lives, and have had a great impacton people’s daily life and socio-economic development. In recent years, Most of countriesincreased road mileage to solve the transportation needs of the people, but the the roadmileage growth rate always cannot meet the growing traffic demand. Only with theintegration of state-of-the-art information technology can we manage the complexity of roadtraffic system. In order to solve the above problems, the intelligent transportation system hasbecome a new research field.Vehicle detection technology based on visual sensor with a highly efficient and accurateperformance is the key elements to the traffic information monitoring of the trafficintersection. Only detect the vehicle accurately can obtain the position information, speedinformation and other information of the vehicle on traffic intersection, then get the roadoccupancy statistics information and traffic flow through time. These facts are basis ofautomatic control of the traffic. In addition, the vehicle detection technology can also monitorthe real-time traffic events and behaviors of the traffic intersection, such as the presence of redlight running illegal parking, over speed and other illegal acts in the surveillance monitoringarea. Therefore the technologies of vehicle detection, identification and tracking records basedon the vision sensors of the traffic intersection has become the core of the intersection vehiclemonitoring technology, and the core technology of ITS technologies. The vehicle monitoringsystem based on the visual sensor of the intersection is important for the traffic informationcollection and emergency events management, which with theoretical and practicalsignificance.This article focus on solving the problems of vehicle detection, identification and trackingbased on the surveillance video sequences which are captured from cameras paved on the cityjunctions and highways intersections. Since the intersection camera is set up at a relativelyhigh position, the occlusion of the identification object need not be considered. The camerasof the traffic intersection are mounted in relatively fixed positions, The shape and size of the vehicle which appears in a fixed area does not change. The cameras are erected at a highheight, so the target vehicle captured by camera is relatively small, The interferences such asvehicle shadows and other confounding factors can be neglected. The vehicles which neededto identify is relatively small, so that the system requires the identification and trackingalgorithms have the high accuracy and robustness features.The main contents are as follows: Firstly, we selected a algorithm with fast processingspeed, small system overhead, high detection accuracy and robustness as the vehicle region ofinterest extraction method for the system presented in this article. Secondly, suitedcharacteristics and classifiers is chosen for the vehicle monitoring system based on visualsensors which are paved on the traffic intersection and classifiers, and then by using thesecharacteristics and classifiers to judge the region of interest of the vehicle. and then weselect the high processing speed and robustness of the tracking algorithm, effective integrationof tracking algorithm can well realize the video-based intersection vehicle tracking. Finallyintroduce a plan which includes designing and combination of detection, identification andtracking algorithm, to form a complete vehicle monitoring system based on video surveillanceon the traffic intersection, and get the traffic flow information, intersection congestion trafficparameters of traffic intersections and highway junctions.After comprehending the vehicle detection, identification and tracking method based onthe visual sensor on traffic junctions, this article will focus on how to integrate these methodsreasonably, and ultimately a vehicle detection system based on the intersection of the visualsensor, and the overall system and each modules of the system were tested by experiments.The vehicle monitoring system based on visual sensors can be used for highway intersectionand urban roads mouth, which have a huge number of vehicles and weather conditionschanging circumstances. The system presented in this paper has broad application prospects.