Co-host: The Institute of Image Electronics Engineers of Japan, Cybermedia Center, Osaka University
In this paper, we propose a method of digesting video, which include many dialogue scenes, by extracting topic and dialogue parts. In this paper, we show a method of extracting topic parts by using the results of camera motion analysis and key shot analysis. Then, we show a method of extracting dialogue parts by using the shot interactivity. Furthermore, we show empirically the effectiveness of camera motion analysis which is prior to shot detection.