Three frame difference motion detection algorithm can easily cause the foreground objects image contour fracture and blank holes because the logical “AND” operations are used between the results of two frame to frame difference with intersections to extract moving objects. In order to improve the algorithm performance, a double three frame difference method is presented, which uses logical “OR” operations between the results of two or more three frame difference with intersections to extract moving objects contour. This kind of union of set operation can reserve the integrality of the extracted contour to a great extent. The concrete logical construction of the new algorithm is constructed on the base of the successive four and five frames images in a video. The simulation results show that the double three frame difference method is effective in improving three frame difference performances.