Video content analysis using dominant sets

Abstract

In this paper, a graph-based method for video content analysis is proposed. The characteristics of the detected shots are investigated for news, commercial, animated cartoon, basketball and documentary videos and experimental studies are realized on these videos. The maximum clique on the weighted and undirected graph, which is constructed according to visual content, is tried being detected. It is inferred that specially in news and commercials, the proposed method can be used for temporal video segmentation.