Research And Implementation Of Vehicle Violation Detection And Tracking Algorithm In Complex Scene

With the economical development and people’s living standards improved,the number ofChinese residents’cars is rapidly growing constantly.Then a series of traffic problems follow,such asurban congestion,vehicle violation and so on,which are widespread concerned by society.To thisend,the requirements for the systems involving urban road traffic management,traffic counts andvehicle violation detection are particularly urgent,especially the vehicle violation detection oncomplex scenes based on video.And the system mianly contains background modeling,motionvehicle dection,tracking and discrimination of vehicle violation.Firstly,several common background modeling and background update methods are discussedin the paper.On the basis,we compare their pros and cons and the Frame Difference is choosed asthe background modeling of urban complex scene in this paper.And then a fast background updatemethod is used on the city crossroads of this paper,which can ease the virtual shadow andincompletely update phenomenon that many current background update methods can’tsolved.Secondly,the common motion vehicles detecting and tracking methods are studied in thepaper.And the tracking method based on Kalman filtering theory is used as the tracking method inthe paper.Thirdly,it’s turn to discusses the shadow issues in the tracking process in the paper.Wemainly study shadow removal algorithm based on texture and cross-correlation function.Finally,themethod of detecting and discriminating vehicle violation is studied.We mainly focus on discussingthe vehicle peccancy line pressing,illegal parking and illegal U-turn. The experiments show that thealgorithms used in the paper are practical and feasible,which can meet the expected designdemands.