Abstract

Superresolution algorithms have demonstrated impressive image-restoration results in the space domain. We consider the limits on superresolution performance in terms of usable bandwidth of the restored frequency spectrum. On the basis of a characterization of the spectral extrapolation errors (viz., null objects), we derive an expression for an approximate bound on accurate bandwidth extension for the general class of superresolution algorithms that incorporate a priori assumptions of a nonnegative, space-limited object. It is shown that accurate bandwidth extension is inversely related to the spatial extent of the object and the noise level in the image. For superresolution of sampled data, we present preliminary results relating bandwidth extrapolation to the difference between sampling rate and the discrete optical cutoff frequency. Simulation results are presented that substantiate the derived bandwidth extrapolation bounds.

© 1993 Optical Society of America

Mathematical extrapolation of image spectrum for constraint-set design and set-theoretic superresolution

Supratik Bhattacharjee and Malur K. Sundareshan
J. Opt. Soc. Am. A 20(8) 1516-1527 (2003)

Superresolution and convergence properties of the expectation-maximization algorithm for maximum-likelihood deconvolution of incoherent images

José-Angel Conchello
J. Opt. Soc. Am. A 15(10) 2609-2619 (1998)

Restoration of band-limited images by an iterative regularized pseudoinverse method

Junji Maeda and Kazumi Murata
J. Opt. Soc. Am. A 1(1) 28-34 (1984)

Analysis of the CLEAN algorithm and implications for superresolution

David L. Fried
J. Opt. Soc. Am. A 12(5) 853-860 (1995)

Superresolution of binary images with a nonlinear interpolative neural network

Carlos A. Dávila and B. R. Hunt
Appl. Opt. 39(14) 2291-2299 (2000)