Infrastructure Software for Mining Image Data Bases

Diamond Eye is a computer program that enables a user equipped with only a personal computer, web-browser software, and a network connection to analyze large collections of scientific image data. The system is based on a distributed applet/server architecture that provides platform-independent access to image mining services. A user interacts with the system through a Java applet interface that is dynamically downloaded when a session is established. There is no need for the user to install "client" software or perform upgrades; the latest stable version of the applet is available automatically. Each server program is typically co-located with a large image repository to enable mining the data in place. Servers are also coupled with an object-oriented data base and a computational engine such as a network of high-perfor mance workstations. The data base provides persistent storage and enables querying of the "mined" information. The computational engine provides parallel execution of the most demanding parts of the data-mining task: image processing, object recognition, and querying-by-content operations. Diamond Eye is currently being used to locate and catalog geological objects in large image collections, but the design provides infrastructure for a range of scientific-data-mining applications. The system can be easily extended to incorporate domain-specific algorithms in any executable form (translation to the Java language is unnecessary).