Study Of Traffic Flow Detection And Statistics Technology Video-based

Traffic flow parameters detection is an important research field of intelligent transportation system. video-based traffic parameter detection with the advantages of rich information, configuration flexibility and low cost, has become a research focus world widely and has a wide application outlook.This paper focuses on vehicle detection and statistics using video processing method. The main contents include:(1) Detection region setting and geometric correction. In the first frame of the video data, the lane is set manually and the lane information is stored; with the lane boundary information, detection area is set vertically to traffic direction; the image in detection area is projected to a rectangular area through geometric transformation.(2) Binarization segmentation of moving object. Traditional methods of target detection are compared and analyzed. Against features of actual traffic scenes, the moving target detection algorithm is combining the methods of frame difference and background difference, is implemented. This part mainly improve the algorithm of frame difference, background updating, and empty and virtual shadow eliminating.(3) Target extraction and counting. Vehicle information data are extracted from the binary image in detection region, the two-dimensional boundary of target region is determined, and according to center of gravity location of the target area boundary, appropriate matching criterion is chosen to determine vehicles and count the number of vehicles.This paper implements a reliable and real-time vehicles detection and statistics, which can be used to realize the precise target detection and vehicles statistics. Shanghai outer ring measured data show that:the algorithm related to this paper can accurately measure the traffic flow over time, and the actual measurement accuracy attains 95%. Experimental results show that the method of traffic flow detection and statistics meet the requirements of vehicle flow detection, and can be used in the traffic information collection system of highway, national highway and city roads of an important region.