Image Analysis in Astronomy for very large vision machine

TL;DR

This paper presents a complex hardware/software system for real-time astronomical image analysis, enabling efficient detection of luminosity variations indicative of new exoplanets using large CCD cameras and distributed computing.

Contribution

It introduces an automatic vision and decision software architecture capable of handling very large images and real-time event discrimination in astronomical observations.

Findings

Enabled on-line detection of luminosity variations

Supported large CCD cameras up to 16k*16k pixels

Utilized distributed parallel network with 256 workstations

Abstract

It is developed a very complex system (hardware/software) to detect luminosity variations connected with the discovery of new planets outside the Solar System. Traditional imaging approaches are very demanding in terms of computing time; then, the implementation of an automatic vision and decision software architecture is presented. It allows to perform an on-line discrimination of interesting events by using two levels of triggers. A fundamental challenge was to work with very large CCD camera (even 16k*16k pixels) in line with very large telescopes. Then, the architecture can use a distributed parallel network system based on a maximum of 256 standard workstations.