Abstract

We may define super-resolution as the ability to form an image with meaningful spatial frequency content at spatial frequencies for which the optical instrument has an optical transfer function (OTF) equal to zero. The 1964 paper by J. Harris¹ showing that no two distinct, finite sized objects could result in the same (ordinary) focal plane image, no matter how slight the differences between the two objects, established a basis for believing that super-resolution was possible—that a signal processing algorithm could be found which could start with the ordinary focal plane image and, building on the details of that image, form a image of the object that clearly manifested all the fine details of that object. After some period of effort aimed at the development of a computationally tractable signal processing algorithm that would allow super-resolution to be achieved,^2,3,4 it was concluded that a prohibitively large signal-to-noise ratio would be required,⁵ and interest in the subject waned.

© 1992 Optical Society of America