Abstract

Prior knowledge of the subband energy partition of an image is useful for recovering an image from insufficient or noisy data. We consider how such knowledge can be associated with constraint sets and derive two algorithms for image recovery. One constraint assumes knowledge of a reference function, while the other assumes knowledge only of a reference energy. The former leads to an iterative convex projection algorithm, while the latter leads to a generalized projection algorithm. Reasonably good recovery is obtained when the irradiance data are not significantly smoothed and other prior knowledge is sufficiently precise.

© 1992 Optical Society of America

Signal recovery with similarity constraints

Hui Peng and Henry Stark
J. Opt. Soc. Am. A 6(6) 844-851 (1989)

Restoration of signals from their signed Fourier-transform magnitude by the method of generalized projections

Eitan Yudilevich, Aharon Levi, G. J. Habetler, and Henry Stark
J. Opt. Soc. Am. A 4(1) 236-246 (1987)

Image recovery from correlations

Timothy J. Schulz and Donald L. Snyder
J. Opt. Soc. Am. A 9(8) 1266-1272 (1992)

Projection-based image restoration

Henry Stark and Elwood T. Olsen
J. Opt. Soc. Am. A 9(11) 1914-1919 (1992)

Convolution-based framework for signal recovery and applications

A. Enis Çetin and Rashid Ansari
J. Opt. Soc. Am. A 5(8) 1193-1200 (1988)