Determination of optimum iteration number for Fourier iterative blind deconvolution algorithm
Takeshi Onodera · 16th Congress of the International Commission for Optics: Optics as a Key to High Technology · 1993
The Fourier iterative algorithm for blind deconvolution, originally proposed by Ayers and Dainty [1], has the advantage of simple structure which allows easy incorporation of various kinds of constraint conditions.Although the algorithm has been shown to be effective for the blind deconvolution of rather simple computer-simulatedconvolved patterns, when it is applied to more complicated patterns such as real blurred pictures, the convergence properties lose stability and it becomes necessary to find the optimum iteration number at which the algorithm is terminated. It is empirically known that EN(psf), the sum of negative values, which are contained in an estimated point spread function for blurring, can be used as an appropriate cost function for determining the optimum iteration number [2].