INTERACTIVE RETRIEVAL RELEVANCE FEEDBACK APPROACH - A TOOL FOR CONTENT BASED IMAGE RETRIEVAL (CBIR)

摘要

Content based image retrieval (CBIR) has become one of the most active research area in the past few years. Many visual features representation have been explored and many systems built. Relevance feedback and region based representation of the image are two effective ways to improve accuracy in CBIR, but still these research efforts establishes the basis of CBIR, the usefulness of proposed approach is limited. One of the reason could be these have relatively ignored two distinct characteristics of CBIR system 1) the gap between high level concept and low level features, 2) subjectivity of human perception of visual content. This paper proposes RF based interactive approach, which effectively takes into account the above two characteristics of CBIR. During this process of interactive retrieval perception subjectivity and the user high level query are captured by dynamically updated weight based on user feedback.