Research On Traffic Signal Control System Based On Machine Vision Technology

By comparing the current traffic signal control system detection methods and traffic flow detection technology,the video vehicle detector has a good application prospects.As a premise,using the method of video signal single-point adaptive control for the vehicle detector as research ideas,this paper studies on the motion compensation mechanism,vehicle tracking mechanism,vehicle detection mechanism,comprehensive judgment mechanism and traffic parameter calculation of video vehicle detector methods,and so on.Besides,this article puts forward the motion compensation method of FAST feature point and RANSAC algorithm,the vehicle tracking method of combining on-line tracking learning with optical flow,the vehicle detection method based on depth learning of RBM,and the comprehensive judgment method based on the Hidden Markov Model.And then,applies these techniques to video detectors,linked to the Simens ST800 signal machine of SCOOT system in order to achieve traffic signal intelligent and real-time control with traffic volume and other traffic parameters at the intersection.In order to verify the effectiveness of these methods,we install devices in the road intersection and test the relevant data.It verifies the feasibility of the methods by the experimental data of traffic signal timing optimization in traffic volume peak period,off-peak period and low period,verifies the reliability of the system by the vehicle detection accuracy at the environment of normal daytime,nighttime,backlighting and rainy weather respectively.From the experimental data,the detection results under these methods in a variety of environments to an acceptable range of signal control.The single-point adaptive control realizes by these methods is very superior in the off-peak and low period of traffic volume,and the experimental data show good results in all aspects,meet the conditions of large-scale application.