This paper deals with the video surveillance problem in static camera. The aim is to design a moving object detection system based on advanced CPU DM6437. After capturing the pictures according to the timing, the background model is established without the objects and the background information is real-time updated due to the changes of the environment. After that, background subtraction is used so as to figure out the moving objects. The morphological methodology and area marking are utilized for eliminating the noise. Experiment results state that the system can identify the moving objects accurately under the error permission.