Researches On Intelligent Traffic Light Based On Machine Vision

With the rapid development of science and technology,various of technologies that have been applied to all walks of life in society,improved the efficiency of labor and promoted the economy make great progress.As an outstanding product of modern technology,especially in recent years,due to market demand,the automotive industry is booming,to bring great pressure on the traffic road.It is why fail to solve the traffic problems with traditional traffic control,resulting in traffic congestion,traffic accidents and other problems occur frequently.Because timing control can not play a crucial role in traffic,it is necessary to establish intelligent system in order to lead traffic flow to drive immediately in effective time.Because timing control and classical genetic control algorithm have themselves disadvantages in the flexibility and convergence speed,the paper propose to optimize the genetic control algorithm.Through experiment datas demonstrate the advantages of the optimization genetic control algorithm in the adaptive,error rate and convergence rate.The paper mainly includes image processing module and adaptive decision module.In the image processing module,in consideration of more than 99% lane line on crossing are straight.After learning Hought transformation extract single lane,based on linear equation is proposed;It is more stable for median method which based on statistical and inter-frame difference to extract background.Use inter-frame difference,binarization and morphological expansion of the vehicle recovery,vehicles’ number would be counted.In the make decision module adaptively,the paper carries out the training of the fuzzy controller through the a priori information,and relevant experience of intelligent transportation transfer into the fuzzy controller,maked the controller have some self-learning ability that adjusting the green-light time of each phase adaptively to lead traffic flow to drive immediately in effective time.In the paper,it is efficient for optimization genetic algorithm to improve convergence speed,reduce the time of searching optimal solution and satisfy real-time requirements.Lastly,average vehicle delay regard as three algorithms’ assessment index.Simulation experiments show that compared with timing control,optimization genetic algorithm’s average vehicle delay time reduced by 33%;compared with the classical genetic algorithm,optimization genetic algorithm’s average vehicle delay time increased by 40%.