How to identify segmented pics Numbers and Letters by comparing them to a set of templates?

Hey everyone, I'm kina new to the whole image processing experience in MatLab. I would really appreciate any sort of guidance or tutorial regarding the stage i'am currently stuck on:

I actually trying to code for a letter and number recognition program, so far i have managed to remove most of unwanted parts in a pic and proceeded to extract all connected components which allowed me to get the segmented numbers and letters of interest.

Coming to the stage that i'm currently stuck on: Having a set of templates for the letters and numbers, i would like compare each template with the segmented numbers/letters and then proceed to display the name of the template that a segmented number/letter matches to. And maybe, if possible to view the percentage of how the segmented pic matches the template.

Once again i would appreciate any sort of guidance or any sort of input on how to proceed to solve this issue.

Thanking you in advance.

Hash L.

Answer 1

[Image Analyst]

See my demo (below the image, in blue text) that uses normalized cross correlation.

[Hash L]

Thank you very much for the demo and sorry for the delay in replying.

So using normalized cross correlation i would be able to compare the best matched template for each segmented pic by finding the brightest spot for each right?

Once again thankyou for taking your time to answer my question.

[Image Analyst]

Yes, though performance degrades as the stuff you're looking for rotates or magnifies/demagnifies.

[Hash L]

I recently came across the corr2 function, which seem to give +1 if the two images in question are the same. So I just wanted to know will I be able to use this as an alternate solution or is this function actually used for some other different situation?

thanks again.