Research On Highway Machine Visionmeasurement Method

With the development of the Internet of things and the popularization of high way visualinspection, machine vision measurement has come into being in the internet of things. As isknown to all, as a consequence of the high speed of vehicles in the high way and the highdemand of real-time, many efficient and precise methods can’t be well applied to the highway visual inspection. In this paper, we put forward the high way vehicle speed measurementbased on machine vision. It is used to do analysis and processing process to the highwayscenario，in order to realize the vehicle speed measurement and provide effective monitoringdata for intelligent transportation. Many moving vehicles may appear in one monitoring fieldof vision once, so the high way visual inspection of machine vision has to monitor and trackevery vehicle, measuring how far they move in per unit time. In the paper, we will try to solvethe problems mentioned above.Firstly, establish the detection methods of high way moving vehicles. In the paper, weanalyze three kinds of motion target detection methods and raise multiscales mixed Gaussbackground modeling method to carry out background subtraction. Sudden illumination,shadows and fog pretreatment are cleared before the moving vehicle detection. And wepropose a new method which treats the fog as dissemination pattern to get rid of the mist.Secondly, further improve CamShift tracking method to adapt to high way mechanical visionmeasurement. Bandwidth down regulated and forward strategy suiting high way is added inthe tracking and optional process. We also use the Kalman predictor to forecast the next flamemotion of the vehicle position. Thus it will suit the automatic and efficient tracking of movingvehicles. Then we also put forward vehicle positioning method for high way vehicle. Themoving vehicle movement distance is calculated by the use of complete geometric locationmethod，to realize the location of moving vehicles.Finally, the proposal is verified by the experiment.The moving vehicle detection andtracking as well as the motion distance movement are verified in the video，we get theposition in the monitored area of moving vehicles through moving vehicle detection and tracking, and obtain the motion trajectory of vehicles in the monitoring period. We make useof the distance measurement method to determine the real distance in a three-dimensionalmonitoring scene. Then corrected FIMI Frame Count is used to get the exercise time, thuscalculate the movement speed of the vehicle. Experiments show that the average deviation ofresults is within4%，which can meet the real-time requirements of highway vehicle speedmeasurement and can be applied well.