What you are referring to is called Optical Character Recognition (OCR). There are a lot of articles on Codeproject on how to do this.
I agree with Zoltán Zörgő, there isn't many free libraries available. However, Tessnet2 is one of the most accurate free engines that you should take a look at if you don't want a commercial engine.
Also, you might want to take a look at Microsoft's engine. See the following link:
How To: Use Office 2007 OCR Using C# (The same link Zoltán Zörgő posted).
Note: OCR engines give the best results for black and white images.