Video Vehicle Tracking And Traffic Information Extraction

The video-based vehicle detection system has developed quickly in recently years, since this approach works better in many ways,such as wider-area detection and easier system setting up, it has become a hot area in Intelligent Transportaion System(ITS). Video surveillance technology provides a direct and convenient way to understand traffic environment. Great importance has been attached to video surveillance technology which is based on video image processing, analyzing, and understanding. Acorrding to the key techniques of ITS, a mothod of video-based vehicle tracking and the mothod is used to exract traffic information and detect traffic incidents have been described in detail in this thesis.The main ideas of this thesis can be decribed as follow:1. For vehicle detection, an algorithm using Non-parametric Model-based has been adopted toextract background. On the research of detection of shadow of moving objects, the paper adopts not only color feature of shadow, but moving vehicle space feature. The following experiment proves its validity.2. The moving vehicle tracking problem is discussed at one condition: the camera is still, amethod using region based multiple target tracking has been improved and realized, Object matching and mutual split in context of multi-objects tracking is discussed in detail. The moving vehicle state forecast linear model in which the system state is characterized by an eight-dimension vector is founded by adopting Kalman filter theory. The matching scope is narrowed and the searching speed is accelerated.3. For traffic information extraction, the result of tracking can be fitted beeline while is a road.Because the road width is fixed value, Several methods are provided to slove some classical problem about vehicle states and traffic incidents detection, meanwhile, using the fixed value, the object detection result can be improved. Especially, the relation between vehicle speed or vehicle shape and the result of tracking have been analyzed and described in detail, then this paper proposed the formula to describe the relation.This thesis provides many ways to resolve the difficult problems of video vehicle tracking andtraffic information extraction system. The results of our experiments show that the methods andalgorithms proposed in this thesis are adaptable.