Multimedia Database Retrieval: A Humancentered Approach

This book deals with adaptive content-based retrieval systems and techniques in image and video database applications. It discusses key considerations in the development of these methods, and demonstrates their performance. It first introduces a perceptually inspired non-linear paradigm in user-controlled interactive retrieval (UCIR), providing an effective tool to bridge the perceptual gap between high-level semantics in human vision and low-level features used by computers. Then, a novel machine-controlled interactive retrieval (MCIR) method is presented to optimize image search in distributed digital libraries over Internet, by minimizing bandwidth requirement in retrieval and reducing human subjective errors. The adaptive methodologies and techniques are then generalized to content analysis and retrieval of video data with time-varying nature, which provide multiple-level access to a large quantity of digital videos. Finally, the UCIR and MCIR systems are demonstrated by applications ranging from Digital Asset Management (DAM), Geographical Database Retrieval, Searching and Retrieval of Art Documents, and Film and Video Retrieval.