Design, Video Pedestrian Motion Detection System Based On Dm6437

Pedestrian detecting and tracking is a very active and practical research in video image processing area. The detection and analysis of pedestrian play an improtant role in the traffic statistics system in metro, crossroad and supermarket, especially in modern urban because of the complex traffic environment. The research of detecting moving pedestrian has great social and economic benefits. In order to realize a more modern, reliable and convenient intelligent detection scheme, a fully automated intelligent system is built using computer vision and image processing to achieve the pedestrian's orientation, identification and tracking through the analysis of the sequence video frame.After doing research of a large number of domestic and international relevant papers, the author found that the research which based on DM6437 platform about counting pedestrian is less. Only a small amount of literatures do research on counting pedestrian in the environment of the supermarket entrances and exits, crossing and bus traffic. However, the system, based on computer environment, has many shortcomings, such as hight cost, poor stability, inconvenient in installation and so on. The environment of this research is the mall entrance and exit. This paper decides to study has certain challenge due to the frequent crowd, goods which are not pedestrian target, and the influence of illumination.The main content of this paper is to study the pedestrian detection and tracking technology based on the sequence frames of video, and the application of this technology to the traffic detection system. This system is completed by using the hardware environment of TMS320DM6437 development board and use TI's emulation software CCStudio_V3.3 programming with C language.The main contents of this paper arrange as follows:First, we study on establishing and updating the background model in the environment that background keeps static. Second, we achieve the motion segmentation and extraction of the moving target area. In the process we add the pretreatment which includes the inhibition on target's shadow and morphological processing for image to improve the result. Third, we achieve the real-time tracking on multi-target, and counting the pedestrian within the valid range of targets area.