A Graph Theoretic Approach for Scene Detection in Produced Videos

Efficient video browsing requires indexing of videos so that the users can quickly locate the segments of their interests. While browsing a video, the users often prefer to skim through the video by scenes rather than frames or shots. In general, a scene can be defined by the continuity in the visual contents of shots due to fixed physical setting or by the continuity of the ongoing actions. We exploit this fact and propose a novel approach for clustering shots into scenes by transforming this task into a graph partitioning problem. This is achieved by constructing a weighted undirected graph called a shot similarity graph, SSG, where each node represents a shot and the edges between the shots are weighted by their similarities. Both color and motion information are utilized to compute shot similarities. The SSG is then split into smaller story units by applying the normalized-cut technique for graph partitioning. The proposed approach is robust and produces meaningful temporal segmentation of video, which is useful for applications such as “video on demand”. Experiments are presented with promising results on two Hollywood movies.