Please see my comment to the question. To start with, please try:
http://www.codeproject.com/search.aspx?q=%28OCR+OR+%22Optical+Character+Recognition%22%29+%28%22C#%22+OR+%22.NET%22%29&doctypeid=1[
^].
However, I cannot guarantee any final results. Even though some works are very impressive; with good quality of code, I haven't find anything ready to use so far (some other members may correct me if I'm just unaware of something, which would be quite possible), all I knew so far would require big amount of high-qualification work.
[EDIT]
See also my past answer and other answers and 7 articles referenced in it:
i need an ocr for a camera image capture[
^].
It's possible that my references are a bit obsolete, but then it would mean luck for you, because it could be because some new works have been added. But don't hold your breath: OCR is in fact a very difficult topic, a remarkable success in it in near past was due to extremely high demand, as then it was a pike in the recognition of literature and other legacy documents. Still, not much of good software was released to open source. One big problem is the quality: many works good from the standpoint of applied mathematics were wonderfully weak in computer science and programming technology, many authors did not really understand such a relatively simple thing as Unicode, and so on…
—SA