Newsgroups: comp.ai.fuzzy,comp.ai.neural-nets,fj.ai
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!oitnews.harvard.edu!purdue!lerc.nasa.gov!magnus.acs.ohio-state.edu!math.ohio-state.edu!howland.reston.ans.net!newsfeed.internetmci.com!hookup!news.caren.net!news.join.ad.jp!wnoc-tyo-news!news.nc.u-tokyo.ac.jp!newssinet!wakasato!rssun3!nakano
From: nakano@rssun3.cs.shinshu-u.ac.jp (NAKANO Yasuaki)
Subject: Re: SUMMARY: Seeking for -> to handwritten numeral recognition
Message-ID: <DEqMtE.Cq4@wakasato.cs.shinshu-u.ac.jp>
Sender: news@wakasato.cs.shinshu-u.ac.jp
Nntp-Posting-Host: rssun3.cs.shinshu-u.ac.jp
Organization: Dept. of Information Engineering, Shinshu Univ., Nagano, Japan
References: <430p17$q9@hrlgw92.hrl.hitachi.co.jp>
Date: Mon, 11 Sep 1995 11:05:38 GMT
Lines: 55
Xref: glinda.oz.cs.cmu.edu comp.ai.fuzzy:5615 comp.ai.neural-nets:26810

In article <430p17$q9@hrlgw92.hrl.hitachi.co.jp> ruck@hrl.hitachi.co.jp (Ruck) writes:
>A week ago I posted an article asking for pointers to papers on
>handwritten digits pattern recognition. I got four responses. Thanks
>again to those who kindly responsed my inquiry. As I promised, below
>is my original post, followed by the responses that I received.

I have not noticed the original post. I went to Canada this August to
attend ICDAR'95, so it might be the reason why I did not noticed it.
I think this article should have also been posted to other groups, such
as comp.text, sci.image.processing and/or fj.comp.image.

From the address of the original poster, I recognized that he/she might
be a HIVIP researcher at Hitachi. I want to point out that Hitachi is
one of the best companies in character recognition in the world. Hitachi
sold more than 10,000 sets of HT-4171 OCR, which can read handwritten
numerals. I think this number is a world record of the OCR society for
a singel model (as a hardware model).

My humble opinion is that the original poster should have asked members
at HRL first, then to those in Hitachi, and then worldwide. From this
article, some readers might recognize that Hitachi has no experiences
and kowledges on character recognition. Of course, a poster can select
his/her way to ask in a company or to ask it at worldwide.

There are many papers on handwritten character recognition, and one might
be able to access to the special issue on OCR in the Proc. of IEEE, Vol.80,
No.7. There are many papers on the subject.

As for other papers, there are the proceedings of ICDAR'93 (Second 
International Conference on Document Analysis and Recognition) held at
Tsukuba and those of ICDAR'95 held at Montreal this August.

Copies of the proceedings of ICDAR'93 might be obtained from IEEE CS
Press. As for a copy of the proceedings of ICDAR'95, I am not sure if
it can be purchased from IEEE CS Press, but it is very probale.

There are workshops dedicated to the handwritten character recognition,
i.e. IWFHR (International Workshop on Frontiers in Handwritten Character
Recognition). The proceedings of the third workshop is published from
North-Holland, as "From Pixels to Features III". The proceedings of the
fourth one, which was held at Taipei in 1994, has not been published as
a hard-cover as far as I know.

As for the recognition program, NIST (National Institiute of Science and
Technology) supplies a program for handwritten character recognition. I
think some netters can follow me on the issue.

The Post-Office Laboratory of Japan (I am not sure if this is correct as
the offical name) held contests for handwritten numeral recognition in
Japan. The Proceedings of the workshops sponsored by the Laboratory were
published, but I am afraid that it might be difficult to get them for
those other than the participants. As for me, I have none of the workshops.

Yasuaki Nakano
Shinshu University
