How does anti-blur software work?

Question

I was very interested by some recent software I discovered to remove blur from a picture. So, I was intrigued how it worked. Of course, it did not work perfectly but did quite a good job of it.

Answer 1

There are some software packages which do this.

Vladimir Yuzhikov's SmartDeblur and the much more complete ImageMagick come in mind as first free solutions but most probably there are other ones.

The most common approach is the one which is the deconvolution based on the Wiener filter

If you want to read more, there are also other methods for achieving this.

From the tone of your question I assume that there exists a small possibility to be a little distant from the mathematical apparatus which appears in the links above, hence I'll try to explain in few, inexact and lay therms what is happening:

A deconvolution method tries to minimize the 'noise' (unwanted information) from a certain data. There are two important things here:

1. How to find what is 'noise' (simply said, is based on its frequency in an image)
2. How to remove it (based on a target mathematical/statistical function which must reach an extreme - minimum or maximum - depending on the method employed)

I hope that I wasn't too scientific.