Content-Selection Based Video Summarization

The paper presents a new framework of video summarization which can extract and summarize the user interested video type from a long video with complicated content. First, the shot detection process adopts the color and edge information to make the shot boundaries more accurate. Then the clustering process classifies the shots according to their similarity of motion type and scene. Finally, we select the important shots of each cluster in the skimming process by adopting shot-importance filter. The filter determines the importance of each shot by computing the motion energy and color variation. The proposed method can produce a classified video summary, which let the user review and search videos more easily.