Future-Viewer: An Efficient Framework for Navigating and Classifying Audio-Visual Documents

In this paper we present an intuitive framework named Future-Viewer, introduced for the effective visualization of spatiotemporal low-level features, in the context of browsing and retrieval of a multimedia document. This tool is used to facilitate the access to the content and to improve the understanding of the semantics associated to the considered multimedia document. The main visualization paradigm employed consists in representing a 2D feature space in which the video document shots are located. The features that characterize the 2D space's axes can be selected by the user. Shots with similar content fall near each other, and the tool offers various functionalities for automatically nding and annotating shot clusters in the feature space. These annotations can also be stored in MPEG7 format. The use of this application to browse the content of few audio-video sequences demonstrate very interesting capabilities.