Design And Realization Of Video-based Surveillance System In ITS

Video-based traffic detection system is traffic surveillance system which draws pretty a lot attention in recent years. Compared with traditional traffic surveillance system, video-based traffic detection system has many advantages like wireless, information abundance, flexible usage, larger detection region, as well as no disruption to the traffic flow because there is no need to dig out the road surface for installation and maintenance. The development in image processing and micro electronic devices has made this detection system available for wide use. On the basis of the project'Intelligent Transportation System'and the survey work of recent technologies in video-based traffic detection system, this paper will discuss the background generation and updating module, and the algorithms of shadow detection.Instead of complex neural network algorithm, this paper will use statistical method to accumulate the static pixels values and produce an initial background for the background differentiation procedure. Meanwhile, we will apply a simple and effective algorithm to update the background dynamically. In the shadow module, classic detection methods take the shadow region as a whole. However, our way will firstly divide the shadow regions into three sub regions with consistent characteristics, and then perform the detection by the grey value reduction rate and the color shift angle. Finally, this method will be refined by shadow edge detection to improve the accuracy.At last, this paper will present the results figure in background generation, shadow suppression and vehicle detection, and compare the shadow miss detection rate and error detection rate with other shadow methods. From the results and comparison, we can conclude that our shadow algorithm has obviously improved the detection accuracy. The research in this paper offers a new way to realize the video-based traffic system and can be the reference for further research.