Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!zaphod.mps.ohio-state.edu!samsung!spool2.mu.edu!news.cs.indiana.edu!msi.umn.edu!noc.MR.NET!uc!shamash!timbuk!bobo
From: bobo@pecan15.cray.com (Bob Kierski)
Newsgroups: comp.sys.mac.hardware
Subject: Re: Scanner as photocopier?
Message-ID: <112848.17440@timbuk.cray.com>
Date: 7 Jan 91 17:36:05 GMT
References: <18490@shlump.nac.dec.com> <959@chem.ucsd.EDU> <1990Dec31.173038.9729@smsc.sony.com>
Reply-To: bobo@pecan15.cray.com (Bob Kierski)
Organization: Cray Research, Inc.
Lines: 26


In article <1990Dec31.173038.9729@smsc.sony.com>, dce@smsc.sony.com (David Elliott) writes:

|> Simple compression (Lempel-Ziv or Huffman) probably wouldn't do much
|> in this case, but what about FAX-style compression?

In a scanned image there is usually a lot of white space so Lempel-Ziv 
compression can save up to 75% maybe more.

|> 
|> Also, how good is OCR these days?  My experiences with this area are 10
|> years old, but they involved very careful circumstances (OCR font belt
|> printers, expensive scanners), and the results were pretty
|> disappointing (scanner operators usually lasted about 4 months before
|> they either quit or moved up).  How easy is it now to recognize, for
|> example, newspaper or magazine article characters scanned at reasonable
|> resolution (high enough to get the dots, not so high that ink artifacts
|> get in the way)?

I have an AppleScan and Omnipage and have been very happy with the performance.
I have been able to read articles in less than 30 second with about 95% acuracy.
As the print gets smaller however, it gets less acurate.
--


Have a day,