Scene Extraction for Video Clips Based on the Relation of Text, Pointing Region and Temporal Duration of User Comments

Recently, video sharing Websites that allow users to attach comments to video clips have attracted much attention. In this paper, we propose a method whereby users can easily retrieve video scenes relevant to their interest. Our system makes both a text and non-text analysis of a user´s comment and then retrieves and displays relevant scenes for viewing of the scenes along with attached comments. The text analysis works in tandem with non-text features, namely, the pointing region and temporal duration of user comments. In this way, our system supports a better organized retrieval of scenes that have attached user comments with a higher degree of relevancy for the user than is currently available with conventional methods, for example, using matching keywords. We describe here our method and the relation between the scenes and discuss a prototype system.