An Improvement of Motion Compensation by Using Camera Position Information
Akihiko SUGIURA, Taisuke KIMURA
This paper describes a new method of global motion compensation to achieve highly efficient video coding. First, we propose a method to extract the pan-tilt vector of an image sequence by using information about the motion of the camera. Second, we study the problem of distortion which is coused by the optical aberration and stands out at the joint of pictures. To solve this distortion, we propose to restore the image using a linear approximation function. The result of computer simulation indicate that the proposed methods reduce the distortion to less than 1 pixel and are effective in improving the picture quality.