A new video retrieval approach based on clustering

The shot based clustering and retrieval is very important for video database organization and access. We can take advantage of semantic clustering to enhance the performance of traditional content-based video retrieval. In this paper, text feature and color feature are first extracted directly from shots. Subsequently, a two-level hierarchical clustering structure that provides indexing scheme for retrieval is constructed by incorporating text and color features. The top level is clustered by text features while the bottom level is clustered by color features. The constructed structure is then used for the cluster-based retrieval. The effectiveness and efficiency of our approach are demonstrated by the experimental results in retrieving and ranking similar video shots.