For fingerprint matching, the data sets will need to have some degree of correlation. Correlation detection has been around since vacuum tube logic systems, so there is a whole lot of background in existence. The trick is to use a "sliding overlay, since different fingerprints may not always start at the same point. That technique is not new either, so you should be able to find references about that as well.
I take it google didn't pop up exactly what you wanted :) Matlab is a great tool for this sort of thing. Most Universities have this tool, and probably have the Image Processing toolbox. You will find methods for image manipulation and correlation. I take it you want to match a scanned image against a set of stored images. Matlab is perfect for this. I think what I would do is develop the scheme in Matlab, and translate to C or HDL myself. Sounds to me like this application cries out for the images to be binarized. An extremely fast and efficient algorithm could be developed around binary correlation. Good luck.
I'm a student industrial engineering. I have to make an embedded design for a fingerprint reader. I have not a lot experience, but i want to make a good design. Can someone give me a good algorithm for fingerprint analysis. I want to calculate: how long it takes to make a fingerprint,the power-consumption, ...