Design And Implementation Of Video Detection System Based On OpenCV

Intelligence Traffic System (ITS) as a special important application of video monitor system in modern traffic domain relieves the gap between high speed increasing of the amount of vehicle and low speed constructing of highway. Vehicle detection system based on video, as the lower layer in ITS, gathers raw information from highway and processed these information be gathered. To be raw information resource of ITS, the accuracy of Vehicle detection system directly works upon to the accuracy of ITS. However, either kind of vehicle detection system in researching step recently has its own leaks:(1) One Kind of Vehicle detection system is just to do the difference between adjacent images. If the difference is up to a certain threshold, the Vehicle detection system considers moving objects bursting into, and then it will send to the plate recognition system a trigger signal. Though the vehicle detection algorithm is simple and quick, but it will produce of a very great amount of false triggerings which will greatly increase data processing burden between Vehicle detection system and Central data processing system and network communication burden between modules and in turn waste a lot of system resources.(2) One Kind of Vehicle detection system is just to do the difference between images and background. If the difference is up to a certain threshold, the Vehicle detection system considers moving objects bursting into, and then it will send to the plate recognition system a trigger signal. Though the vehicle detection algorithm is simple and quick, it has a problem that how to rebuild the background when it has changed by light or other cases.To deal with the outstanding problems above, author’s work is divided into following areas:(1) An algorithm has been proposed which uses high symmetry of vehicle’s contourto locate the sub-windows would like to contain vehicle. In detecting process, first all edges in origin image will be detected, then the edge appeared high symmetry attribution will be located as the interesting region which has high probability to contains vehicle. The experiments prove:algorithm can rapidly detect and locate vehicle and discard most regions without vehicle.(2) the information system of vehicle detection system analyzed. Based on the requirements analyze and main function of the web information system, UML analyzing and modeling are carried out from System user cases analyzing and modeling, to the modeling of activity diagrams, timing diagram, state diagram, deployment diagram, collaboration diagram of the system in a cascade manner. The B/S structure is selected, with respect to the requirements of web information system, for the design of this system. Based on the requirements analyze, system functional module, system class, and database are designed. At last, feasible techniques are selected for the system realization which includes user logon, online video playback, real-time communication and history file management.(3) At last a new vehicle detection system based on the algorithms proposed above has been developed.