Abstract

A fast iterative deconvolution technique that permits the convergence speed of the iterations to be varied is introduced. In this technique, iterations are made to converge as fast as, twice as fast as, and three times (i.e., any integer may be used) as fast as the ordinary methods. The speed of convergence depends on the amount of noise in the data being deconvoluted. This technique is particularly useful for speeding up convergence of the reblurring procedure. The technique converges for all impulse-response function types. The mean-square error versus the deconvolution iteration number for different integral values of the convergence speed of the iterations (1, 5, and 10) is studied for two data sets with and without noise. It is shown that for noisy data sets one has to have control over the convergence speed of the iterations. This technique is also tested with a real data set obtained from an optical multichannel analyzer.

© 1995 Optical Society of America

Iterative deconvolution algorithm with quadratic convergence

C. E. Morris, M. A. Richards, and M. H. Hayes
J. Opt. Soc. Am. A 4(1) 200-207 (1987)

Iterative image restoration for linearly degraded images. II. Reblurring procedure

Satoshi Kawata and Yoshiki Ichioka
J. Opt. Soc. Am. 70(7) 768-772 (1980)

Algebraic analysis of the Van Cittert iterative method of deconvolution with a general relaxation factor

Chengqi Xu, Idriss Aissaoui, and Serge Jacquey
J. Opt. Soc. Am. A 11(11) 2804-2808 (1994)

Convergence of the van Cittert iterative method of deconvolution*

N. R. Hill and G. E. Ioup
J. Opt. Soc. Am. 66(5) 487-489 (1976)

Deterministic convergence in iterative phase shifting

Esteban Luna, Luis Salas, Erika Sohn, Elfego Ruiz, Juan M. Nunez, and Joel Herrera
Appl. Opt. 48(8) 1494-1501 (2009)