Funny, I think me, linuxz and oomlx had a talk about it at #arabeyes.
There has been a paper published outlining some nice methods for arabic OCR, but the math of it is just beyond me.
here is the paper:
http://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-495.pdfOn 9/25/07, Mohamed Magdy < mohamed dot m dot k at gmail dot com > wrote:Salam
As mentioned earlier
http://lists.arabeyes.org/archives/developer/2006/September/msg00013.html
It may be worthwhile and faster if Arabic support is implemented into
Tesseract-ocr ..
The important thing is the support of unicode.. tesseract 2.0
http://code.google.com/p/tesseract-ocr/ can use and understand unicode
and could be trained for any language that don't have its characters
joined..
What it is lacking is mentioned in the training page :
> Tesseract can only handle left-to-right languages. While you can get
> something out with a right-to-left language, the output file will be
> ordered as if the text were left-to-right. Top-to-bottom languages
> will currently be hopeless.
>
> Tesseract is unlikely to be able to handle connected scripts like
> Arabic. It will take some specialized algorithms to handle this case,
> and right now it doesn't have them.
>
http://code.google.com/p/tesseract-ocr/wiki/TrainingTesseract
I did a very very simple test :
http://groups.google.com/group/tesseract-ocr/browse_thread/thread/b1b27838c68681ab
If you could help, please please do so.
Note:- As far as I know, right now..there is NO working Arabic-capable
OCR engine.. free or otherwise.. I doubt if Sahkr software can detect
anything.
--alnokta
_______________________________________________
Developer mailing list
Developer at arabeyes dot org
http://lists.arabeyes.org/mailman/listinfo/developer
_______________________________________________
Developer mailing list
Developer at arabeyes dot org
http://lists.arabeyes.org/mailman/listinfo/developer