Automated cataloging and analysis of sky survey image databases: the SKICAT system

We describe the application of machine learning and state-of-the-art database management technology to the development of an automated tool for the reduction and analysis of a large astronomical data set. The 3 terabytes worth of images are expected to contain on the order of 5 x 10^7 galaxies and 5 x 10^8 stars. For the primary scientific analysis of these data, it is necessary to detect, measure, and classify every sky object. The size of the complete data set precludes manual reduction, requiring an automated approach. SKICAT integrates techniques for image processing, data classification, and database management. Once sky objects are detected, a set of basic features for each object are computed. The learning algorithms are trained to classify the detected objects and can classify objects too faint for visual classification with an accuracy level of about 941Z0. This increases the number of classified objects in the final catalog three-fold relative to the best results from digitized photographic sky surveys to date. The tasks of managing and matching the resulting hundreds of plate catalogs is accomplished using custom software and the Sybase relational DBMS. A full array of scientific analysis tools are provided for filtering, manipulating, plotting, and listing the data in the sky object database. We are currently experimenting with the use of machine discovery tools, such as the AUTOCLASS unsupervised classification program, on the data. SKICAT represents a system in which machine learning played a powerful and enabling role, and solved a difficult, scientifically significant problem. The primary benefits of our overall approach are increased data reduction throughput consistency of classification; and the ability to easily access, analyze, and create new information from an otherwise unfathomable data set