Research Of Vehicle Speed Measurement Through Video Based On Moving Object Detection And Tracking

With the development of computer hardware, network, digital imageprocessing and computer vision technologies, a intelligent monitoring system cansubstitute people to receive and process information and then make the decision. Thisthesis mainly studies how to process and analyses the video frequency on thehighway, and then get the speed of vehicle, provide the speed information for ITS(Intelligent Transport System).This thesis mainly studies two problems: time and distance measurement.Fordistance measurement, this thesis puts forward a new way of image localization basethe detection of the white mark on the road surface. This way can provide theconvenience from the position of something in the image to its real position in theworld. This way doesn't need setting the complicated things for camera calibration;and it is more practicable and adaptable than the line-type. At the same time, thecamera is on a roadside pole, would be better on maintainability and aseismatic, andit is easy to install, adjust and mend. The circulation on the road could not bedisturbed. This thesis puts forward a mark-detecting algorithm base on thresholdvalue division and edge detecting to detect the position of three marks and theirvertex on the highway.This thesis present a vehicle detecting algorithm base on background subtractionand vehicle tracking algorithm base on Kalman filtering, which can track the vehicleon the highway. And we can removal the shadow through shadow detecting algorithmbase on edge detecting. Then we can get the positions of the car in all the images ofvideo frequency. Finally may obtain the track of the vehicles in the image and in thereal world. Calculate the speed of vehicle according the image localization and thetime-gap of video. A vehicle speed measurement system through video are designedin this thesis.This thesis analyses the problem of distance measuring and moving objectsdetecting and tracking, and puts forward my ways to solve them and test themthrough experiments. The result from the experiments gives a positive answer to the accuracy and stability of the algorithm, and these way meet real-time andpracticality requirements.