The Reach On Intelligent Video Surveillance System Based On ARM System

As an important element of secure and defensive system, video surveillance technology is undergoing a rapid development. Intelligent, low power consumption, easy customization, low-cost video surveillance system is a focus of current reach. The study subject of this thesis is the research on intelligent video surveillance system, which is based on the technology of embedded system. And an embedded video processing system with the capabilities of motion detection and segmentation is designed and built.Firstly, this thesis introduces the back of video surveillance technology at home and abroad, and then presents a new distributed framework of intelligent video surveillance system. By this framework, the realization of machine vision is devided into two steps. This thesis put research focus on the first step, in which motion detection and segmentation are the main tasks.In this thesis an embedded video processing system is designed to do the intelligent works of the first step. This system is composed of hardware and software. When it comes to the hardware, an embedded multimedia development platform named"Z228"is selected to be the core of the system, which contains ARM9 processor, MPEG-4 hardware codec and many control interfaces.Then design of the other models of the system is also described. On the software part, we choose Windows CE Operating System (OS), and describe its customization for our system, as well as the design of drivers for the other models in our system. The algorithm of motion detection and segmentation is the key of our system, so, we proposed a new improved background model with its updating method for embedded system, The algorithm first establishes an background based on Gaussian model; second computes the set of statistical parameters of background model according to the way pixel changes; and then extracts moving objects using the updated background model.Finally we apply the algorithm to the embedded video processing system, and test it in a variety of environments. The experimental results show that the whole system takes on nice stability and can achieve accurate segmentation result in real-time. At the same time, the system is robust to some abnomal conditions, which are common in real surveillance.