Multichannel Blind Deconvolution As A Tool For Solar Image Restoration

Abstract

A principle of restoration methods based on multichannel blind deconvolution (MBD) is introduced. The methods assume that for every un-degraded unobservable image several degraded observed images are available. It is better conditioned than classical single channel approach. The first algorithm represents a generalization of iterative deconvolution scheme introduced for single images. The second MBD algorithm is based on so-called subspace technique. The subspace method is not iterative and this possibly implies an implementation that can be computationally more efficient. Both methods are presented in applications to artificial image data (computer-generated multichannel degraded data) with known ideal image to get a comparison with restored one. Performance in a real situation on solar photosphere images is shown.