Aspects of handling data from astronomical images

Abstract

Astronomical photographic data bases can have sizes ranging from a few thousand to many billions of bytes. Much of the data processing can be regarded as a filtering process which reduces these large input sets to very much smaller subsets. We give as an example the problem of detecting a complete set of astronomical objects of a specific type (QSOs), and describe in detail the algorithms, data structures and processing techniques used in the SRC Automatic Plate Measuring machine at Cambridge. This system completely analyses 4 billion bytes of data per day.