Webview: A Distributed Geographical Image Retrieval System Wei Wang, Daoying Ma, Yimin Wu and Aidong Zhang Department of Computer Science and Engineering State University of New York at Buffalo Buffalo, NY 14260 USA David M. Mark Department of Geography State University of New York at Buffalo Buffalo, NY 14260 USA URL of demo: http://picasso.cse.buffalo.edu:8080/META/ Abstract With the increasing number of geographical image databases on the Internet, it is a very important issue to know about the most relevant image databases for given user queries, in order to achieve high efficiency in retrieving the images. To address this problem, we have developed a novel system, Webview, which intelligently ranks the distributed geographical image databases based on their visual content. Our system summarizes the visual content of each database in a metadatabase, which is a collection of automatically generated image templates and statistical metadata. With the metadatabase, our system can guide the user queries toward the most relevant image databases in a distributed environment. 1 Introduction Geographical images are being gathered from civil, defense, and intelligence satellites at an explosive rate on the Internet. One of the challenging problems for content-based image retrieval (CBIR) systems is to retrieve the relevant images from remote image databases given user queries. Under such a distributed environment, it is impractical for the system to examine all the images in the databases to match for the relevant images. Instead, based on a particular visual query, the system should be able to rank the relevant image databases intelligently by the metadata gathered and summarized from respective image databases. The query is then guided to the respective databases and the relevant images in those databases are returned to users. Based on our previous works on the data resource selection in distributed visual information systems (Chang et al, 1998) and on CBIR (Zhu et al, 2000; Sheikholeslami et al, 1999), Webview is devel- oped to support content-based retrieval on the distributed geographical information system (GIS) image databases. In the following sections, we will first introduce the system architecture of our system, then focus on the design and implementation of the metaserver armed with the metadatabase and the database ranking mechanism, and finally describe the user access applications. 2 System architecture The system includes three major components. The first component is the remote GIS image databases and respective image retrieving servers. The second component is the metaserver, including the meta- database, the metasearch agent, and the query manager. The third component is the user access applica- tions at the client machines. The components are illustrated in Figure 1(a). This research is supported by the National Science Foundation Digital Government Grant EIA-9983430.