An iterative SNR estimation algorithm for wiener deconvolution of self-similar images distorted by camera shake blurring
Agustı́n Marcelo, Jacoby Daniel · 2008
The Wiener deconvolution technique can successfully restore real world images distorted by camera shake blurring. However, its performance is highly sensitive to the estimated SNR, a parameter that is hard to estimate correctly a priori. An iterative algorithm is proposed to find a suitable SNR by gradually testing the Wiener deconvolution on a fragment of the blurred image and observing the self-symmetry of the result, a quality measure that can be estimated by comparing the vertical and horizontal gradient distributions.