Motion object detection is the basis of intelligent video surveillance;the speed and the detection accuracy of the existing algorithms need to be improved.By analyzing the luminance change between adjacent frames

this paper presents threshold calculation method based on the third central moments of frame difference

and proposes an anti-noise background maintenance model and finally we combine the ideas of background subtraction with frame difference algorithm to detect motion object.Furthermore

morphological post-processing method is introduced to reduce noise and improve detection precision.Experiments on standard sequences and outdoor/indoor sequences

the algorithm has the advantage of high-speed

anti-noise and it can detect the obstacle motion object.