What do you need to accomplish ?
What platform are we talking about ?
If for windows, and if you have Microsoft Office, then you already have MODI.
http://www.codeproject.com/KB/office/modi.aspx
(problem with this is, that if it is for distribution, all people may not have MS Office)
Then there's tesseract, a multiplatform open source OCR currently developed by Google (really dunno how much they develop, mostly looks like they just want to spam their name, by "taking over development" as it seems they arent doing much to it), but people got it working fine (no thanks to google).
http://code.google.com/p/tesseract-ocr/
You may also wanna check out GOCR
http://jocr.sourceforge.net/
But if you are a coder/scripter, it should be possible to use ImageMagick for OCR, the "learning" way, teaching your application what the characters looks like, and then compare your stored taught characters with the image containing the text you want to extract, alot of work, and would probably be awfully slow, but it could be done.