Moving objects detection is an important part of intelligent video surveillance system ,its aim is to Split moving objects from background image in video sequences. Until now, the main algorithms of moving objects detection are background difference method and inter-frame difference method[2][3]. the background difference method can obtain moving objects information relatively complete, while rather difficult in obtaining and updating the background model. inter-frame difference method has strong adaptability.,but the target speed must be moderate. And inter-frame difference method is sensitive to the noises, but difficult to segment moving objects completely. In order to obtain better detection effect ,the paper adopts the improved algorithm of updated background model of inter-frame difference by combination of background and inter-frame difference.