Kalman Filter and Double Step Warping based Image Prediction

M. Pieper and A. Kummert (Germany)


Image prediction, image warping, KALMAN filter,computer generated image sequence


The main aspect of this paper is the presentation of a new method for increasing the frame rate of computer generated image sequences. Based on image warping the new frames are predicted only on basis of the viewpoint trajectory, the previous frame information and the corresponding z-buffer content. The future viewpoint coordinates and its orientation are tracked in two steps by means of a KALMAN filter (KF)[1]. The two prediction steps are spaced by a quarter of the old frame period. Based on the viewpoint parameters two frames are predicted. The source of the second frame is the result of the first prediction step.

Important Links:

Go Back