Index-based fast search algorithm of image database on Internet

For application over the Internet, a multimedia system should efficiently manage various types of data such as text, image, audio, and video. In addition, the rapidly increasing number of users is making network bandwidth more and more precious. In this paper, we propose a novel algorithm for fast image database searching over the Internet. In the proposed system, the search results first returned return rough images and the user can preview and narrow down the images he wants. By interactive feedback, the user can obtain more accurate results by refinement. We present some experiments to measure the efficiency of the proposed algorithm. In addition, the quality of the query result is about 1 dB better than the JPEG coded image at the same bit rate. Based on these results, our search algorithm is an effective retrieval method for image databases