The Research Of Moving Vehicles Tracking Technology Based On Machine Vision

Machine vision technology through the simulation of human eye function,makes survey, judgment and analysis to the real world,its birth brings the very big convenience for people's work and the life,it also pave a new road for many domain research and application.With the problems of transport highlights,Intelligent Transport Systems(ITS) provides a system solution to solve the traffic problems,the machine vision technology has already became an important research direction and the inevitable trend in the ITS.The study of moving vehicles tracking problem in the ITS based on vision has been become a hot spots recently.How to detect and track the vehicles from the complex traffic video background with real-time and accurately is the present research difficulty and the hot sport,the paper launches the discussion and the research to the question.Aiming at the large-scale,complex context of the moving vehicles tracking problem, through analysis and summary of the current moving target detection and tracking methods based on vision,first propose the correlation improvement measure in the part of detection, such as first strengthens images and then processing method,selective statistical background updating algorithm,the synthesized algorithm of improving background subtraction and symmetrical difference method and so on.It provides more accurate information of the parameters for the following tracking link.In the moving vehicles tracking part,the paper uses advantages and features of the tracking methods based on model,and combines with the physical characteristics of moving cars,sets up the cube model for moving vehicle,and proposes a matching algorithm based on model.In the paper,we introduces the current common math treatment methods of the moving targets tracking,and the characteristics of the method which is suitable for the moving objects. According to non-linear characteristics of moving vehicles,the paper introduces EKF algorithm into the tracking link and forecasts the moving informations,and proposes a tracking framework based on the cube model and EKF algorithm,it also presents in detail for the implementation algorithm based on the cube model and EKF.The simulation results of the above algorithm show that the proposed method in this paper can obtain better detection and tracking effects,and provide an effective data source of moving vehicles for tracking study and application.It provides a solid foundation in solving problems of ITS and the further application.At the the mean time the paper can also provide guidance and reference to others moving target applications of detecting and tracking based on machine vision.