Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!uwm.edu!cs.utexas.edu!uunet!willett!ForthNet
From: ForthNet@willett.UUCP (ForthNet articles from GEnie)
Newsgroups: comp.lang.forth
Subject: Machine vision
Message-ID: <525.UUL1.3#5129@willett.UUCP>
Date: 22 Feb 90 01:57:08 GMT
Organization: Latest link in the ForthNet chain.  (Pgh, PA)
Lines: 34


 Date: 02-19-90 (17:24)              Number: 447
   To: ALL                           Refer#: NONE
 From: BOB JENNER                      Read: (N/A)
 Subj: CHARACTER RECOGNITION         Status: PUBLIC MESSAGE

 I'm looking for guidance, references, etc. about character
 recognition.  I'm considering a job whose principal technical nut
 to crack is identifying parts by scanning 1 to 3 digit numbers
 printed on them.
 .
 The numbers are printed in block letters roughly 1/4 inch high,
 on background zones of different, and generally contrasting
 colors.  On some parts there is some, but not a lot, of
 background clutter in the form of lines cutting behind the
 numbers, or color changes behind the numbers.  Variation in
 placement of the numbers on the parts can run to 1/8 inch or so.
 .
 The (prospective) customer would like to do this at 5 per second.
 The parts will be stopped during the scan, though.
 .
 The font may vary from production run to production run, but will
 be constant during any given run.
 .
 Anybody offer recommendations in the way of references, hardware
 and / or software?  Or a sanity / reality check: is this pushing
 the state of the art, or is it pretty do-able?  Is there anybody
 who has done something similar and who might be interested in
 consulting on the character recognition?

 NET/Mail : LMI Forth Board, Los Angeles, CA (213) 306-3530             
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: 'uunet!willett!dwp' or 'willett!dwp@gateway.sei.cmu.edu'