Content based Video Retrieval, Classification and Summarization: The State-of-the-Art and the Future

This chapter provides an overview of different video content modeling, retrieval and classification techniques employed in existing content-based video indexing and retrieval (CBVIR) systems. Based on the modeling requirements of a CBVIR system, we analyze and categorize existing modeling approaches. Starting with a review of video content modeling and representation techniques, we study view-invariant representation approaches and the corresponding performance analysis. Based on the current status of research in CBVIR systems, we identify the video retrieval approaches from spatial and temporal analysis. Subsequently, we present the video classification approaches from multidimensional distributed Hidden Markov Models. Finally, a summary of future trends and open problems of content-based video modeling retrieval and classification is provided. 1 Video Content Modeling and Representation In this section, we give a generalization of video content modelling and representation. We first investigate the general problems in video content modelling and representation. With regard to that, we present the state-of-the-art approaches: curvature scale space (CSS) and centroid distance functions (CDF)-based representations. We subsequently propose the null space invariant (NSI) representations for video classification and retrieval due to camera motions. Moreover, we propose the tensor null space invariant (TNSI) representation for high dimensional data. Finally, we give a brief overview of the other approaches in video content modelling and representation and future trends.