I had developed an OCR which was completely dedicated to font type & its size. I just used to calculate hight & width of the font & find out the ratio which I compare with the datebase value & then choose the font.
I want to extend the logic to all type of fonts & sizes. I want to do it using Artifical Intelligence. I have seen the Coding of some of the open source OCR. But, it won’t be useful.
Please guide me the way I should follow/start to developer my own OCR. I want to make it as an Oper Source Project for whole world & also want to provide API for futher project specific developemt in Java.
Thanks in Advance.
There are a lot of algorithms and concrete tools for that purpose. Mainstream trend is using artificial neural networks for this. In java the best ANN library is Neuroph. You can also use Matlab for this and then call it from java (there a lot of tutorials how to do this http://www.cs.virginia.edu/~whitehouse/matlab/JavaMatlab.html
There is also R languge which can help you.
Also there are plenty of new approaches for OCR now: Computational Intelligence or even such things:
http://www.amazon.com/Knowledge-Based-Intelligent-International-Computational-Intelligence/dp/084939807X