Video is a medium that is suitable for recording dynamical aspects of the real world. For obtaining information from a video, the operation "play" is generally used. This operation provides an intuitively understandable presentation for users, but it has a drawback that it requires users to take a long time for an observation. This is because developments of browsing mechanisms that supports efficient observation of video content characteristics have been expected. This paper introduce a method for composing abstract presentations of a video in databases for browsing based on semantic content descriptions. The facility of abstract presentations composed by our method is to enable users to grasp an overview of a few hours activity recorded in a video. The format that we adapt for abstract presentations is the storyboard, in which representative frame are layouted on 2-dimensional grid. In this method, semantic contents of a video are modeled as a sequence of events, and the importance of event is defined based on the dependency among events in the same sequence. The calculation of the importance uses statistical characteristics on event occurrences as knowledge, and reflects viewpoints of users. Storyboards are composed based on importance.
|Translated title of the contribution||A Method for Generating Storyboards Adaptable to Video Database Browsing Based on Semantic Content Description|
|Number of pages||15|
|Publication status||Published - Jul 15 2001|