Research On Multilane Traffic Flow Detection Technology Based On Video

As an important traffic information in the intelligent transportation system,traffic flow can directly reflect the traffic condition.Therefore,it is very important to use the computer vision technology to detect the traffic flow effectively to establish a better traffic management system and reasonable urban planning.Based on the theory and technical research of motion vehicle detection and traffic flow statistics,ViBe algorithm for motion vehicle detection and virtual loop-based traffic flow detecting algorithms are analyzed in this thesis.On the basis of above algorithms,a new improved ViBe algorithm combined with adaptive virtual loop for multi-lane traffic detection is proposed.The main contents of this paper are listed as follow:1.The ViBe algorithm is improved to solve the "ghosting" problem in the traditional ViBe algorithm.First,according to the original algorithm for background sample initialization,and then use the idea of image binarization,taking into account the pixels are repeatedly judged as the prospect may be the background sample needs to be updated,the current frame image through the traditional ViBe foreground detection and then To determine the number of times the foreground target to determine the number of times.Finally,change the sampling factor in the original algorithm background update mechanism,and increase the background update probability of the "ghosting" elimination period.Experiments show that the method can eliminate the "ghost" in time and improve the accuracy of the detection of sports vehicles in different scenes.2.The multi-lane traffic flow detection method based on adaptive virtual loop is mainly for the traditional virtual loop can not accurately detect multi-lane raffic flow.Then the fixed detection area opened up,according to the moving target trajectory to establish or cancel the mobile virtual loop,and then use traffic flow statistics algorithm based on virtual loop to achieve traffic statistics.Different scenarios were chosen for compare experiments in this thesis and the experiments show that the proposed method solves the limitation of virtual loop algorithm for multi-lane traffic detection,and the accuracy of traffic flow detection is higher than that of virtual loop algorithm in the case of no reduction in time efficiency.