|
Paper Abstract:
|
A concept of a new software architecture for astronomical image processing is presented. Current approaches of
astronomical data extraction are explained: photometry, astrometry and high-resolution imaging. The methods seem
incommensurate with each other and suffer from systematic errors caused by known issues. Hints are given, where and
how to consolidate these algorithms into a more sophisticated approach to improve data retrieval from images. A detailed
discussion of user experience and possible expectations is presented. The analysis yielded current astronomical image
processing software based on a shell oriented task collection designed and implemented since the 80ies. From a historical
perspective, requirements for any next generation software design can be identified as backward compatibility, platform
independence, extensibility, a more intuitive user experience, documentation issues, and long-term reliability. A smooth
software transition is proposed. Object-oriented and XML driven approaches may help to combine the existing software
packages and future extensions within a common user interface. A unified, formal task description language for image
processing is proposed to support generality. |
