Research On Intelligent Traffic Light Based On Image Processing

With the development of economy and technology,the manufacturing cost of vehicle has decreased,the purchasing power of ordinary people has improved,vehicle have become more and more popular,and have entered more and more families.This not only brings convenience to life,but also causes traffic congestion in major cities.Traffic pressure cannot be well alleviated and current traffic problems cannot be solved by the traditional timing traffic control strategy.Therefore,in order to effectively control urban intersections and improve road traffic efficiency,it is necessary to establish intelligent urban traffic lights which can adaptively change traffic timing according to the real-time change of traffic flow.This thesis mainly divides into two modules: vehicle detection and statistics based on image processing and traffic lights management research.In the image processing part,how to detect and count vehicles is discussed.Firstly,based on video acquisition and corresponding preprocessing,the advantages and disadvantages of different moving object detection algorithms are analyzed and compared,then the background subtraction is chosen as the moving object detection algorithm.Then,different background modeling algorithms in background subtraction are further studied.ViBE algorithm is selected through analysis and comparison with experiments.At the same time,aiming at the problems of ghost detection and shadow detection in ViBE algorithm,an improved method is proposed.Finally,the virtual coil method is used for vehicle statistics.The simulation results show that the proposed algorithm has high accuracy and can be used in vehicle detection and statistics.In the part of traffic lights management,a two-stage fuzzy control scheme which intelligently converts expert experience into knowledge base of fuzzy rules to obtain the period of traffic signal is presented.This scheme changes the fixed period in traditional traffic lights management and determines the signal period according to traffic flow and main road.But also,the phase time can be regulated according by the traffic flow.A multi-objective optimization model with the objective of average waiting time,average parking times and waiting queue length is proposed.Particle swarm optimization algorithm is selected to solve the model.Finally,in order to verify the effectiveness of the traffic lights management method proposed in this thesis.SUMO,a traffic simulation software is used to build multi-intersection road,traffic flow and signal model.Taking Nanjing Central Road as example,sections with different number of intersections are selected for simulation analysis.The simulation result shows that the proposed traffic lights management method can effectively improve traffic congestion and the improvement effect is obvious when the number of coordinate control intersections is three or four.