Abstract-In this paper, we propose a method to estimate the velocity of moving object from a series of images with respect to the model of it. This model-based measurement is calculated by the Kalman filter algorithm. For an object, if one of its model can roughly be found, we show that the estimation based on Kalman filter algorithm is more accurate and faster than any other methods without a model. In this approach, the model of image sequence dynamics is made up by Distributed Parameter Systems (DPS). By using this method, we can get not only translation but also rotation parameters of the moving object. Some simulation examples will be given to demonstrate the use and validity of this method.