A spatio temporal method which aims at segmenting moving objects from stationary background automatically and accurately was proposed. Three steps are followed steps in the algorithm. First, the current frame is segmented into connected regions of uninterrupted grey scale. Secondly, The intersections of each two neighbor defference images are gotten to judge which pixels are moving pixels. Finally, those moving regions are segmented from the stationary regions according to a certain rule. The experimental results show that the algorithm is effective in the segmenting moving objects from the stationary background accurately.