In this study, a method to digest a TV drama based on the mental side of the picture content is proposed and implemented. The video structure model which consists of the tracks of an actor's utterances, BGM, offset noise, effect sounds and video shots is defined and the relation between the chosen frames in the digest and their structural features in a live drama video and the broadcast digest made by the professional beforehand is clarified. The system, which implements the relation, determines the candidate frames for the digest of the video based on the relation information and makes a digest of the specified length.
展开▼