Ocr in c++

mlong

That's a pretty ambitious project to pick up "in a couple of days." I'd suggest you start small and look through the fine Qt tutorials and examples out there, get a good grasp of C++ and the Qt framework, then work up to what you'd like.

Qt doesn't have native OCR support, so it's not just a matter of wrapping up some native functionality into a pretty interface overnight.

r.bhardwaj

thank you sir for the reply.......
but i want to ensure myself that after learning Qt i will able to make OCR using it.Because i have limited time for this project and i have to choose a correct start.

JohanSolo

You can do whatever you want with Qt, that is not the problem. You'd better wonder first how you would achieve OCR. A first guess would be neural network, ... but I'm no expert.

andre

Well, I would start reading up on the ORC engines that are out there already. Unless you want to start from scratch? In that case: not a chance that you'll make it within a year or so, and probably quite a bit more before you have anything useful.

You cannot reasonably expect to proficient enough to tackly a complicated task like OCR while taking shortcuts. You really need to know the basics before you can do advanced stuff. At least, tell us something about your background knowledge. Do you have a good grasp of C++ already? Other programming languages perhaps?

r.bhardwaj

i m not talking about making a full featured ocr. I will satisfy if within 3 months a will become able to make a basic application that can read character from scanned image.
As for as C++ is concerned although i have not made any big project in c++ but i am a regular c++ programer from last 2 years & i m aware about the basics of c++ .
I am reading thoroughly all the steps for making an OCR so plese tell me any library or other thing so that i should start to imlement basic of digital image processing in Qt.
Thank you!

andre

You will probably want to use the "Tesseract-ocr":http://code.google.com/p/tesseract-ocr/ library.

r.bhardwaj

thanx sir for reply........
i have satarted hands on Qt but i m getting problem regarding Tesserect-ocr library . i have searched alote to find some tutorials regarding using Tesserect-ocr library but i m only with its documentation which is very hard to understand 4 a beginner. So please guide me regarding this.

Kretikus

Look up the documentation of QImage. That is the support Qt will give you :)

msx_br

I have used Tesseract with Qt and OpenCV (2.04):

Look at the dlltest project that comes with Tesseract.
You can make a .so ( or .dll ) to use in your project

My approach ( which have achieved best results ) is preprocess the
desired image with text, and pass only the ROIs with the text.

The result was good but far below commercial packages...

But, if want to implement your own OCR engine, you must have a good knowledge in all of the following technologies:

1 - Image segmentation ( OpenCV ??? )
2 - Feature extraction ( OpenCV ??? )
3 - Good algorithm for feature grouping
4 - Text line detection
5 - Character chopping
6 - Character recognition ( see Neural Network for Recognition of Handwritten Digits in codeproject )
7 - Dictionary based context sentece composer ( QRegexp have good support for this )
8 - Context analyzer

mikelilin

This post is deleted!