OT: scanning (was: Re: scanning)
Fairlight
fairlite at fairlite.com
Fri Oct 29 06:35:57 PDT 2010
On Fri, Oct 29, 2010 at 06:06:38AM -0700, after drawing runes in goat's blood,
Bob Rasmussen cast forth these immortal, mystical words:
> The Windows module (twain32.dll) is a common interface to the Twain
> drivers supplied by the various scanners. Talking to twain32.dll is like
> talking to the GDI to do printing. There are still a zillion details to
> deal with.
He's dealing with C code...should be par for the course. :)
>
> The interface itself is rather clunky, and has to be integrated with the
> calling application at a very low level, in order to provide event
> feedback. Every scanner is different. There are many possible features in
> the scanner, so there's some discovery that has to go on. The standard is
> vague in some areas, and some scanners don't appear to follow it
> precisely.
That's a shocker. :)
> There's a significant shortage of independent information about
> programming Twain. The official documentation of the standard is short on
> details.
This explains why TWAIN captures from various scanners and cameras seem to
fail 80%+ of the time when I tried them. I use other methods of getting my
data nowadays...mfgr-supplied scan software and SD cards or USB transfers.
> Most manufacturers of scanners in the home/small office space have focused
> on providing a master scanning application for photos and documents. But
> many of the features supported at that level, such as optical character
> recognition or conversion to PDF, are not supported at the Twain driver
> level.
OCR is a nightmare to even think about coding. Even SnagIt's text capture
is a failure on, say, a browser window or a terminal emulator. Loses or
gets lots of characters wrong. Works perfectly on Windows Explorer or
other Windows-generated windows, but I think it accesses the objects
directly then...or handles system fonts better. Likely cheats...never seen
it mess up a directory listing.
> You might think I'm trying to scare you away from producing a free
> scanning product because I have one I'm trying to sell you. You'd be
> right ;-)
No trouble to me...it's Brian's gig... I've just tried to be helpful.
To say it's a non-trivial project to code scanning and/or OCR is
understatement, and I know it. Anyone with a clue should. I think your
market share is safe. :)
mark->
--
Audio panton, cogito singularis.
More information about the Filepro-list
mailing list