Currency recognition using template matching

I am trying to do currency recognition using image processing. What I am trying is to take a ROI from the paper currency scanned image and use it as a template to compare it with the original paper currency image. In this case I will have the templates for each paper currency and then I compare each template with each paper currency image. I have cropped the template from the original image (I have the ROIs). After doing the template matching with correlation, I got the coloration coefficient matrix. My problem is to compare between these matrix to denominate the currencies. Any one can help me in this or suggest me a different way of approaching the problem?

Just for more clarification of my problem:

I have 1,5,10,50 and 100 unit currencies. I have the scanned images of these currencies. And I have also the templates for each currencies which I cropped form the currency images. Now I am trying to match the templates with the currency images using template matching. For example , in matching the template of the 1 unit template with each currency note images (1,5,10,50 and 100 units images), I got five correlation matrix. My problem is to compare between these five correlation matrix to identify the one which matches with the right currency note, in this case the 1 unit template.

[Ajesh Singhvi]

Hi...I am doing the same project....having the same problem can you help me regarding this pls..

[Jegi]

Please refer my published paper with a title “Automatic Recognition and Counterfeit Detection of Ethiopian Paper Currency” http://www.mecs-press.org/ijigsp/ijigsp-v8-n2/v8n2-4.html