M. Pieper and A. Kummert (Germany)
Image prediction, KALMAN Filter, 3D Object Tracking, image warping
The main aspect of this paper is to present an alternative to increase the frame rate of computer generated image se quences. The static background objects of the new frames are predicted by means of a warping algorithm [1] and the moving objects are predicted and fit into the background frame by means of a 3D object tracking algorithm (3DOT). The image warping is done based on the information of the previous frame, z-buffer, the camera position, and orienta tion. In addition to the camera information, the 3D object tracking algorithm requires the index-buffer, vertex-buffer, and the world-matrix of each moving object to predict the next position of the corresponding object and to map the textures. After the prediction, the new frame is inserted into the image sequence to increase the frame rate without doubling previous images.
Important Links:
Go Back
