didn`t know if it`s the right category, but

I would like to know if/how it`s possiible to read text from an image like it does the [url=www.abby.com]Abby fine reader[/url]...

Meaning not to read any kind of text, but just like this one:
www.voodoo.ema.lv/example.png

I just know that it`s something to do with OCR or neural networks..

Can someone explain it to me? Or give some examples?

If think this might be diffucult, because graphics like the one on your image are used to protect text from reading it out by a computer program.

Visual Confirmation Systems (like maxk said are used for protection against automated registrations to ensure bigger security of a particular service.

To read out such a text you could use either weighted pattern matching (check if the gifen image part matches a known pattern and how likely that match would be) or the way FineReady does by OCR, where OCR would be the better, but more complicated way. To use OCR you need a well trained "NEURONAL NETWORK" that is trained to character detection.

All in all that issue is quite complicated and there exists no generalized solution (I even know some VCS that are even hardly manageable by humans).

Anyone who is capable of being elected president should on no account be allowed to do the job.
Ich code EdgeMonkey - In dubio pro Setting.