Efficient video summarization leads to facilely exploring video content appropriate to the user’s intention with low cognitive demand.In this paper,we present a novel approach for summarizing videos in the form of multi-scale structures that exhibit different video features at different scale levels and allow exploration of video contents with multi-scale interaction.The semantic relationship between structures is addressed and user intention is also considered and integrated in the summarization and interaction.This paper first introduces the concept of multi-scale structures for summarizing video content and describes three different types of structures that present important features at different scale levels.Furthermore,a continuous zooming interaction for browsing multi-scale structures is provided to facilitate video browsing.Finally,an elaborate user study is conducted showing that user performance on understanding and browsing videos is improved.
展开▼