Design Of Traffic Data Collection System Based On Video

With the rapid development of social economy and the improvement of urbanization, the traffic problem is increasing seriously day by day, the requirement for intelligent management of the transportation system increases continually. As an important part of intelligent transportation system (ITS), traffic data collection provides information sources for traffic management, traffic control and forecast, transportation guide, traffic information services and so on. In recent years, DSP technology, embedded technology, traffic information detection technology and video image processing technology has improved significantly. The method of traffic data extraction based on video has been further developed and widely applied in ITS due to its unique advantages.Based on Davinci technology platform of TI company, this paper designed a traffic data collection system based on video. The system can collect real-time traffic video image, realize real-time monitoring of site of traffic. the test parameters of this system can be set intuitively on the computer, it can real-time detect moving vehicle and extract traffic flow, average speed and vehicle occupancy etc. It has vital significance to make road traffic unobstructed and the traffic management intelligent.In this system, front cameras capture multiple traffic line images in real-time, then send the images to video detection processing system for digitization, the final effective traffic data which is extracted by data processing would be transferred to traffic data management center. Video detection processing system is built by TMS320DM6446 processor chips which is dedicated digital video framework based on DSP and ARM dual-core. This chip is proposed to solve problems such as vehicle detection accuracy and real-time limit in software and hardware of the whole system of the past.Design can be divided into two parts as arithmetic design in DSP core and control design in ARM core. DSP core mainly realizes video detection algorithm processing, this paper puts forward an improved algorithm of vehicle detection based on frame differential method, using double virtual test area, realized image difference by comparing the neighborhood, detected the vehicle by correlation between successive frames, greatly improved the vehicle inspection accuracy and real-time. ARM core mainly implements the function modules, The video collection application program based on V4L2 video driver is designed to realize image acquisition, Network transmission based on TCP protocol is realized by socket programming, configuration software used in the detection of inspection parameters settings is designed in VC environment of the PC, Effective traffic data extraction method is designed to realize basic traffic data collection.The result of system test analysis shows that this system has higher detection accuracy, good real-time performance and expansibility. By adopting of the high performance and low power processor chips absolutely lower system cost, this system has good practical value as an important sub-system of ITS.