ALGORITHMS FOR IMAGE PROCESSING OF CROWDED FIELDS IN ASTRONOMY

Abstract

This thesis deals with the image processing of so called crowded fields. These objects (such as Globular or Open stellar clusters) in astronomical images represent extremely dense and concentrated stellar fields and they are extremely difficult to analyse. This thesis brings detail analysis of this phenomena together with the description of the problem and suggestion of several new approaches and algorithms. The presented WHIDE algorithm (Weighed and Histogram Thresholded Deconvolution) deals with crowded fields by using both known (mathematical deconvolution, aperture and profile photometry, Nelder-Mead optimization, ...) and new approaches. The research about the application of standard image processing methods in astronomy and other fields is presented together with simulations and the analysis which helps in designation of the new algorithms. For these simulations new simulator of astronomical images (including instrumental errors and defects) with crowded fields called GlencoeSim is used. The suggested WHIDE algorithm uses as well two new approaches in data reduction - ”Flux Histogram Noise Thresholding” and Statistical weighing of deconvolution results. These methods allow the deconvolution process to be automatic and reliable and deal with both the noise and the deconvolution artefacts.