[bksvol-discuss] Re: reasons for quality disparity
- From: "Sarah Van Oosterwijck" <curiousentity@xxxxxxxxxxxxx>
- To: <bksvol-discuss@xxxxxxxxxxxxx>
- Date: Mon, 13 Dec 2004 17:21:28 -0600
Scan quality depends on many things. I think OCR engine is the most
important facter after the effort put forth by the user to make sure
everything is complete. That doesn't mean a lot of effort is always
necessary, but sometimes it is. The scanner being used is a facter. The
font and layout of the book can make a scan impossible or effortless and
everywhere in between. I can't stand reading a book that isn't in good
condition, so even if I were to only scan for myself books would be
satisfactory to most people. If a book really doesn't want to come out
right I just give up. The number one reason for that is a margine problem
that makes it impossible to get the book flat enough on the scanner. Most
other problems can be optimized away either by manual effort or scanning
package tweaking. Sometimes the font is so weird it doesn't work, or they
have used book content that OCR software hasn't yet been made to deal with,
such as screen shots and diagrams, or odd characters ranging from
mathmatical symbols to other languages, only some of which will be supported
by any one program.
Then there is the problem of people who don't set up the software correctly.
Sometimes that is because of a lack of interest and/or effort on the
scanners part, and sometimes it is just that the person hasn't had the time
or opertunity to learn how to use it. This list can help with the second
problem, but isn't going to solve the first one. All scans can be improved
with the help of a human editor, and many can be improved with the help of
an automated one.
Sarah Van Oosterwijck
http://home.earthlink.net/~netentity/
Other related posts:
- » [bksvol-discuss] Re: reasons for quality disparity
