A moving object dedection-based image sabilization algorithm was propoed for addressesing the moving object problem of image sabilization.In the algorithm,the estimate global motion paramaters were used to compensate the inter-frame global motion,and the difference image between the motion compensated previous and current frames was obtained.Then,the difference image was divided into a series of forground and background blocks.And the global motion paramaters in the next run were estimated only by tracking the background blocks.Simulation results show that the proposedalgorithm has low computatiomal complexity,and achives quite high interframe transformation fidelity,which is suitable for stabilizing video clips with large moving objects.