Interactive visualization of video content and associated description for semantic annotation - Springer

摘要

In this paper, we present an intuitive graphic framework introduced for the effective visualization of video content and associated audio-visual description, with the aim to facilitate a quick understanding and annotation of the semantic content of a video sequence. The basic idea consists in the visualization of a 2D feature space in which the shots of the considered video sequence are located. Moreover, the temporal position and the specific content of each shot can be displayed and analysed in more detail. The selected features are decided by the user, and can be updated during the navigation session. In the main window, shots of the considered video sequence are displayed in a Cartesian plane, and the proposed environment offers various functionalities for automatically and semi-automatically finding and annotating the shot clusters in such feature space. With this tool the user can therefore explore graphically how the basic segments of a video sequence are distributed in the feature space, and can recognize and annotate the significant clusters and their structure. The experimental results show that browsing and annotating documents with the aid of the proposed visualization paradigms is easy and quick, since the user has a fast and intuitive access to the audio-video content, even if he or she has not seen the document yet.