Multi-Gaussian Deconvolution of Elliptical Sources: Reliability Tests Using Ground-Based and HST Images of m32

Abstract

Deconvolution of spherical sources from the Point Spread Functions based on multi-Gaussian expansions (RMG method) is here extended to elliptical objects by using bivariate Gaussians. To test accuracy and robustness of this approach, we have first derived the nuclear structure of M32 using observations obtained by the Palomar 60" under mediocre seeing conditions and by the Hubble Space Telescope (Faint Object Camera and Wide Field/Planetary Camera images before refurbishing). Secondly, real and simulated HST images were also processed by three other restoration algorithms much more challenging from the computational point of view (constrained algebraic minimization, Richardson-Lucy and Maximum Entropy Method). The comparison of the results obtained from data characterized by very different resolutions and using several approaches to the deconvolution proves that the RMG method (i) so far is the most simple and versatile procedure when dealing with sources of regular shape and (ii) no matter how high the resolution of the data, it can improve the extraction of the information content.