Abstract
Conventionally, the video mosaicing works according to the number of image frames in the video sequence. However, the conventional video mosaicing causes two problems. Firstly, the merged image cannot be of high quality because commulative error often occurs on processing numerous image frames. Secondly, the image mosaicing cannot work efficiently when there are many image frames that capture the same region of the object. To solve these problems, we propose the efficient video mosaicing method using virtual camera's 3D point tracking. The virtual camera's 3D points are used for classifying the image frames in the video sequence and selecting the best image frame in each cluster for reducing comulative error and obtaining the high quality merged image. In this paper, the concept, algorithm and experiments are described.
