Abstract

We report on some early testing of a maximum-likelihood-based image-restoration algorithm which has been adapted from nuclear-medicine imaging. Potential uses Include noncoherent imaging systems where, for example, the object being imaged is a self-luminous source. This broad class of systems includes certain types of astronomical and infrared systems, although this presentation focuses on the use of fluorescence microscopy. The testing that is reported has involved both computer simulation of superresolving image restoration and experimentation with the restoration of a real image obtained from a defocused camera. This testing has involved a qualitative assessment of restored images, a quantification of mean square error in the restored Images and in the restored Fourier transforms of the images, and a quantification of resolution In these images. The potential for superresolution and bandlimited-spectrum extrapolation has been shown in these simulation tests, while the defocused-camera experiment has given some preliminary evidence of feasibility with real data.

© 1988 Optical Society of America