ABSTRACT
Deconvolution is an image restoration technique which aims at undoing, literally “unrolling,” the blurring eect any imaging system has on the “true image,” the object distribution. In doing so, deconvolution is a natural companion of superresolution imaging. If a particular image could be deconvolved perfectly one would obtain boundless resolution, and it would not even be necessary to build complex microscopes. Unfortunately, the gain which can be achieved by deconvolution is limited. One obvious source of this limit is that most optical systems entirely stop transferring information about the object beyond a certain amount of detail. It would seem obvious that deconvolution techniques cannot “roll back” information which is not present in the image, but as will be shown in favorable conditions this boundary can be overcome, albeit within limits.
