Research And Implementation Of Traffic Flow Video Detection Based On ARM

With the increase in the number of cars, the problem of traffic congestion is becomingmore and more serious, the establishment of intelligent transport system is considered to bethe effective way to solve the problem of congestion. Video detection technology is efficient,accurate, it has an important position in intelligent transport system. The traditional videodetection technology relys mainly on the computer for processing. In this process, it isrequired for long-distance data transmission, so we need a lot of equipment cost. At the sametime, the data will be distorted and delay in the process of transmission. With thedevelopment of embedded system, embedded processor has powerful function. Theembedded products with high stability, low power consumption, low cost has become anindispensable part of our life, therefore, the vehicle flow detection system based onembedded platform has become the focus of ITS research.This paper mainly studies the traffic image by road camera for real-time acquisition. Tocalculate the traffic flow of the information in the image, discussed issues related to theforeground objects extraction, the shadow removal, data transmission. At the same time, thispaper also discusses the current popular embedded system and embedded processingequipment. At last, we realize the algorithm that based on the embedded Linux system andARM11microprocessor. At the same time, we design a set of traffic statistics system withconvenient and maintenance.This paper mainly includes the following work:1. In the detection of moving vehicle target, We compared the advantages anddisadvantages of the pavement marking method, the two frame difference method, The threeframe difference method and the background difference method. According to thecomparison results, we choose the background difference method. We choose the ViBe basedon background difference method as the prospect target system for vehicle detectionalgorithm. This method is simple and precision.2. In the shadow of moving vehicle detection, We explain the reasons for the formationof the shadow and show the effect of shadow to foreground object. In HSV color space, wemake experiments on the shadow detection and the results show that the detection method isaccurate. At the same time, The quality of image will be reduced during the process ofshadow removal. In order to solve problems of the image quality, the paper also discusses themorphological processing of the image of two value.3. In the vehicle counts, we compared the region labeling method and virtual detectionline method. At last, we the virtual detection line method that is real-time and accurate.4. We study on the characteristics of common embedded operating system, we choosethe embedded Linux system and ARM11processor as the core system and the core processorsystem. 5. We study on the programming tool of ARM processor and complete the design ofApplication program. At the same time, we also complete the acquisition of image,processing of image, display of image, data transmission and other functions.6. We complete the design of program of single chip microcomputer. The function of theprogram is to receive the data and display.7. In the actual traffic on the road, we conduct a comprehensive test of the system. Thetest results show that the system is real-time, accurate, stable. We has achieved the anticipatedtarget requirements.