Design Of ARM-based Vehicle Detection System Of Electronic Police

Many common problems such as traffic jam, road safety, etc. have become concerns around the world. In order to solve such traffic problems, Intelligent Transportation System (ITS) appears. As one of the most important subsystem of ITS, Electronic Police System plays an important role in correcting the acts of drivers, putting an end to traffic accidents, raising the level of executing traffic law, improving the road capacity, etc. And as the key technology of Electronic Police System, vehicle detection technology directly impacts on the performance of Electronic Police System.Presently, vehicle detection technology is based on induction coil in practical Electronic Police System. It is very inconvenient in installing, debugging and maintenance. So vehicle detection technology based on video causes many researchers'concern. Video-based background difference algorithm is usually used in moving objects detection system, which lies on background acquisition and background refresh with mass data computing. So the real-time characteric is not good enough. And the false rate and undetected rate are higher, because of the complexity of vehicle objects. This paper puts forward a vehicle detection improved method with virtual coil. Such method processes less data, because the object area is negligible comparing with the whole image. It makes the real-time characteric higher and the false and undetected rate lower, reducing the cost of installation and maintenance at the same time.From the system hardware cost point of view, the hardware architecture of such vehicle detection system is made up of ARM9 (AT91RM9200) general processor and Xilinx low-cost FPGA (XC2S100) in this paper, ARM9 configs FPGA in slave-serial mode. Real-time image collection is controlled by FPGA and image is processed by ARM. This vehicle detection system could detect vehicles which violate red-light, cooperating with traffic signal controller. Such detection system is developed by embedded technology, and the design of software is based on embedded Linux operation system.This paper builds the experiment platform. From the result of experiments, this vehicle detection system well performed in real-time capacity and veracity. The rate of Image processing reaches 10 frames per second, and the rate of accuracy is nearly 92%. It will be significative that applying such detection system to Electronic Police System. Comparing with video detection sysem based on DSP, ARM-based detection system absolutely lower system cost because of the architecture of cheap ARM9 and low-cost FPGA.