| Video files are usually very large. Displaying a short digital video clip on the Web requires transmission of large amounts of compressed data. When the scene to be visually described is still, a panoramic image can be used to achieve the same visual effects. However, the data size of the panoramic image is much smaller than the video data for the same scene. Therefore, the technique of panoramic imaging may provide an effective solution for displaying large still scenes on the Web.; The existing panoramic imaging methods have severe limitations on imaging conditions and computational complexity. Most of the algorithms also require much user input and interaction. The disadvantages have limited the techniques for practical use.; In our research, we develop a new algorithm for creation of panoramic images from ordinary panned video clips. The algorithm is characterized by automated image creation with very little user input. It does not require any prior knowledge about the cameras used and video-taking processes. It is a robust technique that may produce good panoramic images from video clips of different quality levels. |
