Re: Arabic OCR info

[nadim at arabeyes dot org (Nadim Shaikli)]

On Wednesday - 3.March.2004 [10:26:00 -0500],
 Ahmed Elgammal (elgammal at cs dot rutgers dot edu) wrote:
 > > 
 > >  1. To your knowledge, are there any Open Source Arabic OCR
 > >     solutions available ?  If so, can you please provide a
 > >     URL and some contact info ?
 > 
 > NO.
 > 
 > >  2. To your knowledge, are there any resources we could use
 > >     (papers, books, published algorithms, etc) to implement
 > >     something on our own ?  Any leads would be helpful.
 > 
 > YES.. various publications. Speaking about myself, here are
 > two publications:
 > 
 > A. Elgammal and M. A. Ismail, "Techniques for Language Identification
 > for Hybrid Arabic-English Document Images", IEEE 6th International
 > Conference on Document Analysis and Recognition, Seattle, Washington,
 > U.S.A. September 10-13, 2001. 
 > 
 > A. Elgammal and M. A. Ismail, "A Graph-Based Segmentation and
 > Feature-extraction Framework for Arabic Text Recognition", IEEE 6th
 > International Conference on Document Analysis and Recognition, Seattle,
 > Washington, U.S.A. September 10-13, 2001. 
 > 
 > You can find some citation to others work in these papers.. 
 > 
 > Looking into recent ICDAR conference proceedings would be useful.
 > There are also some survey papers (old though- nothing recent)
 > 
 > >  3. To your knowledge, is there a group or an effort that we
 > >     can get involved with to attain an Open Source Arabic OCR
 > >     solution (given one doesn't already exists) ?
 > 
 > don't  know
 > 
 > >  4. To your knowledge, which of the various Open Source OCR
 > >     solutions (that do not include Arabic support) is the most
 > >     advanced and is worth pursuing and augmenting (from the
 > >     likes of GOCR, ClaraOCR, etc).
 > > 
 > >       http://www.linux-ocr.ekitap.gen.tr
 > 
 > I don't know any details about the technical approach that are used in
 > these open sources. for sure Arabic has unique features that need to be
 > addressed by any algorithm. please see above publication for what I
 > think is different about Arabic writing.
 > 
 > -- 
 > Ahmed Elgammal
 > Assistant Professor
 > Dept of Computer Science 
 > Rutgers University

Thank you _very_ much for your reply.  We will augment (Mr M.Alkarouri ?)
our OCR Wiki with the info noted (hopefully we'll be able to find the
publications noted on the internet somewhere),

  http://wiki.arabeyes.org/ArabicOcr

Mr. Elgammal, in passing would you be interested in assisting us
make this a reality (strictly as a hobby - for fun) ?  I think
it would be of interest to all involved (and the Arabic Open Source
community) to have both academia and volunteers working in unison
assisting each other.  In short, when the time becomes more
appropriate (we are still considering whether to launch an OCR
project due to the daunting task that it is), would you be willing
to supervise and/or lend some "sanity checks" to what we will
potentially pursue ?

Thanks again and Salam.

 - Nadim