S. Aoyagi, T. Takada, K. Sato, T. Sugawara, and R. Onai (Japan)
Multimedia tools, video browsing, video skimming, timereduction
There are now various kinds of video contents available via cable TV and satellite. The usual way to check on video contents quickly is to fast-forward through them. How ever, the sound is not synchronized with the images and is either absent or partial during fast-forwarding, so it is hard to obtain the important information from audio data and we must concentrate on the images to understand the contents. In the near future it will be difficult to select suit able video contents from among the huge amounts avail able in a reasonable time. There have been several studies about video summarizing, but they concentrated only on video segmentation and constructing and representing the meaningful structure of a video, and did not examine about the total playing time of the summarized video. In most of those systems a user must specify the segmented scenes he/she wants to see. In this paper, we propose a new time reduction method for video skimming which focuses on the overall playing time. This method can generate a shortened video of less than a user-specified length without specifying the scenes to play back. We also describe a prototype system and the results of a questionnaire survey about it.
Important Links:
Go Back
