There are books about OCR that can tell you about current techniques. However, since OCR is rooted in computer vision and artificial intelligence, you might want to look to those fields for a starting point.
When you're trying to get a computer to recognize anything, it's usually helpful to do whatever you can to simplify the problem. When trying to recognize objects in an image, you might apply a filter to increase the contrast, another to remove as much noise as possible, and a third to detect edges. A good book on computer vision will explain all that and more. From there you might use the edge information to detect features and try to match those features to forms that you already know about. At the same time, you might use statistical information to reduce the set of likely candiates (e.g. if the last two letters you recognized were 't' and 'h', the next one is probably a vowel or 's'). These techniques come from AI.
If you're really interested in implementing an OCR system of your own, you're going to need to know something about AI and computer vision first.