Methods for video content analysis are necessary for the growing amount of video information delivered to consumers today. In this paper, we present a system for video content analysis called Vitamin, which provides management of a home video library. The system presents the user with a visual table of contents that provides an overview of the video content and direct access to particular points in the stored video. In this process, we apply a computationally inexpensive, simple, yet powerful mechanism for cut detection and keyframe filtering. Our initial implementation and results show that this system can perform video content extraction in real time on a low-end platform that matches a visual table of contents extracted by an expert.
展开▼
