In message  Paul Goodwin writes:
> The problem with deconvolving confocal images is that the signal-to-noise
> is generally poor so that the noise, which can look like small points to
> the deconvolution algorithm tends to get enhanced as well. There are
> methods that have reported good success with this (the blind
> deconvolution people in Holland) and we are working at ways of masking
> the noise frequencies out of our images and OTF's. So far the image is
> improved, i.e., there is resolution extension, but I would be happier if
> the noise was less.
 
AutoQuant (Troy NY) has a blind deconvolution package that can handle confocal
images, too.