The Research Of The Traffic Flow Analysis Algorithms Based On Video Detection Technology

With the china urbanization processing, the constant improvement of transportation system has become the core issues of social development, and intelligent transportation system based on computer vision technology is an increasingly important. Statistical techniques to implement traffic flow using video image processing technology in intelligent transportation systems research process has become an important issue and key technology in the social studies. Implementing traffic flow statistics in real life, you first need to accurately extracted moving vehicle from the background, and then combined with the knowledge of the video detection technology and image processing, to identify the vehicle number, vehicle type, vehicle motion characteristics and other traffic information from a certain range of traffic video images. In intelligent transportation systems, real-time control of the transportation system and optimize traffic can be achieved through the rational use of these traffic parameter information.This paper begins with a brief overview of the basic research of intelligent transportation systems, the development of video detection technology in home and abroad, and several popular traffic flow statistics system; outlined the important of computer image processing technology in the traffic flow statistics; through concluding these detection technologies, show the advantages of video detection systems in intelligent transportation systems. Then discuss the traffic flow detection system based on virtual detection circle in detail. In the process of research focused on the moving vehicle target extraction method based on the dynamic context in highway and propose the improved hybrid Gaussian target detection combined with Bayesian model and traffic flow statistics algorithm fusion of frame difference method and area method.The main work and innovation points are as follows:(1) According to the influence of highway vehicle movement characteristics and weather factors, propose background extraction algorithm of block mixed Gaussian, on the one hand reducing mixed Gaussian computation, on the to improve the detection effect and avoid noise and debris pollution. In the moving vehicle detection process, the introduction of Bayesian model detects movement vehicles, reduce false alarm rate and the missed detection rate.(2) Based on the above background extraction and movement of vehicles detection algorithm propose new adaptive background update algorithm, can be real-time updates background, and improve the efficiency of the detection.(3) According to the framework of traffic volume statistical algorithms, set up double virtual detection line to reduce the factors of vehicle division and too small target.Based on the virtual detection line, constructs traffic flow technology system of fusion frame difference method and area method, and improve the accuracy of detection.In computer operating system, through the programming to realize traffic flow statistics, and test a large number of video of the actual road, achieved good results.