Program in Media Arts a Sciences

Digital imagery has become commonplace because the World Wide Web has been grow-ing rapidly. In order to facilitate global sharing of digital imagery, the goal of this work is to build a Web-based system, Kuafu, for image annotation and search. Annotation and search in Kuafu is based on the content of images. In this work, a cognitive search model is proposed, and the visual and semantic content of images are distinguished. The cognitive model explains that people begin to search using semantic content, and find the right picture by judging whether its visual content is a proper visualization of the semantics given. Kuafu emphasizes semantic content in particular, but utilizes visual content as well. Enabling users to query efficiently is a central role of an interface for image retrieval systems. The disadvantage of the query-by-example paradigm, which is used in most image retrieval systems today, is that the user has to have examples at hand before searching. In contrast, the query-by-design paradigm is proposed, where the user is able, first, to semantically express what s/he wants and, then, gradually specify visual conten