Fast Image Indexing and Visual Guided Browsing

Managing the ever rapidly increasing image data is a very challenging task, which can be roughly divided into user side and computer side issues. Although fundamental problems on the computer side of the challenge, such as image content representation, similarity matching and intelligent retrieval have received extensive research effort, equally important and difficult issues on the user side, such as fast indexing, interactive browsing and system usability have not yet been seriously studied. In this paper, we present a framework that integrates fast image indexing and user friendly interface design for easy and intuitive image database browsing. We use natural scene statistics to divide the database into easily describable visual themes, which in turn render extremely efficient indexing, and intuitive and visually meaningful user interface design. We present computer simulation results to demonstrate the effectiveness of the proposed method.