|Japanese page is here.|
Personalized video digest
In recent years, network-based music and video streaming content are becoming popular, using broadband Internet connections as ADSL and CATV become widespread. Video streaming is expected to be one of the killer applications in this broadband era. At the same time, due to the spread of portable devices such as PDAs and especially cellular phones, Internet use with cellular phones rather than computers is increasing. The next-generation of cellular phone will be supported video streaming services and the distribution of video content, based on the demands or profiles of users. Therefore, it is desirable to create a system that can generate and distribute video digests according to each user's profile. In this project, we are studying an authoring system which adds metadata indexes describing the content to the original video, along with a digest generation system to create video digests from this metadata and the users' profiles.
The following figure shows the flow of the video digest generation using the user profiles. The original video source, such as an MPEG or AVI file, will be prepared for distribution to portable devices. In order to generate the video digests, MPEG-7 metadata describing the original video source is created for the MPEG-4/7 authoring system. This metadata is used as an index to the content of the original video. At the same time as the metadata is prepared, an MPEG-4 video file for video digest distribution is generated. At the other end, in the digest generation part of the system, the importance of each scene is calculated using the metadata. The relative importance is affected by the user profile stored on the user's portable device. The final output is a personalized video digest reflecting each user's preferences.
|Last modified 26 Feb. 2002|