METHODS OF ZERO SPACE FOR BLIND IDENTIFICATION OF POINT SPREAD FUNCTION IMAGE

Abstract

It was shown that the left side null space of the autoregression (AR) matrix operator is the lexicographical presentation of the point spread function (PSF) on condition the AR parameters are common for original and blurred images. The found PSF was used as initial one in Luce-Richardson (LR) iterative deconvolution schema. It was offered amplified variant of LR schema which has fast convergence. The initial PSF was not changed sufficiently by LR iterative process. This fact points on its optimality due to analytic properties. The deblurring of degraded by defocusing, moving, vibration and haze images shows simplicity and effectiveness of the proposed method.