Abstract

The general problem of reconstructing two unknown functions from a number of measurements of the intensity of the Fourier transform of the product of the functions, where the functions are displaced relative to one another by a different amount for each measurement, is examined. It is shown that this problem is equivalent to a blind deconvolution problem, which suggests that a unique solution exists for functions of any number of dimensions. Such a problem arises when attempts are made to recover the specimen and probe functions from microdiffraction plane intensity measurements in a scanning transmission electron microscope. An iterative algorithm capable of recovering these functions is described, and examples of the operation of the algorithm are presented for simulated data involving one- and two-dimensional functions.

© 1993 Optical Society of America

Blind deconvolution of a Hermitian and a non-Hermitian function

Nobuharu Nakajima
J. Opt. Soc. Am. A 8(5) 808-813 (1991)

Reconstruction of phase objects from experimental far field intensities by exponential filtering

Nobuharu Nakajima
Appl. Opt. 29(23) 3369-3374 (1990)

Parametric blind deconvolution: a robust method for the simultaneous estimation of image and blur

Joanne Markham and José-Angel Conchello
J. Opt. Soc. Am. A 16(10) 2377-2391 (1999)

Phase-retrieval and intensity-only reconstruction algorithms for optical diffraction tomography

Mohammad H. Maleki and Anthony J. Devaney
J. Opt. Soc. Am. A 10(5) 1086-1092 (1993)

Surface reconstruction from photometric stereo images

Kyoung Mu Lee and C.-C. Jay Kuo
J. Opt. Soc. Am. A 10(5) 855-868 (1993)