Abstract: An detection algorithm for abandoned an moved objects used in intelligent surveillance video is presented in paper.It based on Gaussian mixture model,using different update rate cumulative mask update background model to establish dual background model.Use edge matching algorithm to identify and extract foreground object,give the space location of abandoned or moved objects and alarm. The experimental results show that this algorithm significantly reduces the computational complexity,the abandoned or moved objects detection with high detection accuracy and robustness,reduce the false alarm rate.