Abstract

An iterative algorithm developed by Ayers and Dainty [1] for the blind deconvolution of two functions is applied to the problem of phase retrieval. Because its structure closely resembles that of the Error Reduction (ER) algorithm commonly used for phase retrieval [2-4], the performance of the Ayers/Dainty (AD) algorithm is compared with that of ER. Both of these algorithms are compared to the faster hybrid input-output (HIO) algorithm [2-4] for the cases of real, nonnegative objects with known and unknown support using Fourier intensity data with different levels of additive Gaussian noise.

© 1989 Optical Society of America