Single-frame blind deconvolution by means of frame segmentation

Abstract
For restoration of an image degraded with a shift-invariant point-spread function (PSF), partial images segmented from the degraded image are blind deconvolved in parallel for estimation of the common PSF in each partial image. An object function is reconstructed by deconvolution of the whole image with the PSF. A technique for calibrating contamination caused by the segmentation is developed. Results for a computer-generated image and an atmospherically degraded solar image are presented.