Representing a video by a set of key frames is useful for efficient video browsing and retrieving. But key frame extraction keeps a challenge in the computer vision field. In this paper, we propose a joint framework to integrate both shot boundary de
