Blind Deconvolution Using the Sure-Blur Criterion and Linear PSF Expansions

The article investigates a new approach to blind deconvolution, which approximates the underlying point spread function (PSF) as a linear combination of a series of PSFs. The coefficients in the linear expansion are the free parameters that we solve for to estimate the true PSF. Minimization of Stein’s unbiased risk estimator (SURE) is used as the criterion to determine the coefficients. The numerical optimization procedure for estimating the coefficients is fast and relatively simple due to recent work [1]. The series of PSFs in the expansion are angled anisotropic Gaussian distributions generated with random parameters, and this approach is shown to work reasonably well in both simulated and real examples.