What you are referring to is called Optical Character Recognition (OCR). There are a lot of great articles on CodeProject on how to OCR. Some of them talk about free methods but from my experience, they either have low accuracy or do not provide many features, such as:
http://www.codeproject.com/Articles/238114/Realtime-Webcam-Sudoku-Solver (does a lot more than OCR, but the OCR part is specific to numerical digits in Sudoku puzzles)
OCR With MODI in Visual C++
(uses MS control in C++, but can use same control in C#).
On the other hand, there are some commercial toolkits that provide more features and higher accuracy. You can find an article on one of them in the following link:
OCR with LEADTOOLS: The Better Choice