Programmer Vladimir Yuzhikov has released a Windows only application SmartDeblur that can do some really impressive things with removing blur from images after the fact. Released on his website earlier this month, the open source project can recover a pretty spectacular amount of information from an otherwise illegible photograph.
Yuzhikov spends a substantial amount of time on the technical details of the mathematical processes he uses to get this information, and if it's something that you understand, it's interesting reading.
Yuzhikov isn't the only one working on deblurring images, and tweaking areas that are otherwise out of focus. Last year, Adobe showed off something similar at their MAX conference, and here's something similar from the Chinese University of Hong Kong.
While the examples of SmartDeblur don't seem to do much to make an aesthetically pleasing final image, what is really impressive is just how much information can be extracted from the fuzzy original. As you can see in the example below, legible text was easily reconstructed.
Yuzhikov has been active in a recent Reddit thread discussing this software, and it is a work in progress. Since it's open source, others are free to try and make it work better. Yuzhikov only released a compiled version for Windows, but others have been attempting to make it work for Mac (with mixed results).
And while SmartDeblur may not be perfect, it shows that in a couple of years, we're going to be able to do some incredible things with images that up until now we would have just thrown out.