I haven't been in the market for an OCR device in quite
a long time so I actually wonder that myself. I imagine that the
next one I get will be my own personal device as the last one I
had was through my job. It was a pretty good system back in 1993
and worked well until it finally wore out about a decade later.
It used a flat-bed HP scanner and software from a now
defunct company called Telesensory Systems Incorporated.
I got to compare it to the Kersweil Reading Edge system
of about the same time period and both systems had their good
points.
If the OCR solutions offered for the IPad/IPhone world
are really good at the OCR function, this is a game changer
because there are no moving parts in the system like there were
with the flat-bed scanners.
Unless the page one is scanning is very simple and in
good condition, OCR systems get confused and misread text. Human
beings, for that matter, sometimes get confused and misread text
so that is not just a problem with machines.
The camera must get a very good picture of the page. The
page must be straight and not slanted in the captured image and
the OCR software needs to be smart enough to not only decode the
print properly but handle layout such as columns and tables in
such a way that a person who is blind can make sense of it.
A lot of stuff is very artistic these days, and you
might hear the text but since it is trying to read in a linear
fashion and the page is maybe laid out in columns or some other
non-linear manner, you will hear bits and pieces that are all
jumbled up in a way that will just drive you crazy.
Straight linear text such as what you have in a story
book or history textbook usually comes out just fine on any
decent OCR system but they are apt to choke on complex pages
like newspapers and magazines.
I am sure there are others on this list who are more
familiar with the cutting edge and have actually tried some of
these newer systems. I would like to know, myself which ones are
reasonable and which to not even consider.
A sort of humorous example of what happens at times
occurred when I got my new OCR system in 1993. At that time, OSU
still shipped tons of paper-based handouts to staff and
students so I used my OCR system a lot for reading mail as well
as computer manuals which, in 1993, were still mostly
paper-based.
The scanner was a black-and-white model which was fine
most of the time as I sure didn't care whether the images were
monochrome or in brilliant vibrant colors.
Actually, I learned to care because the lamp that the
scanner used to aluminate the page was a mercury-vapor lamp
which is a brilliant blue-green like a lot of mercury-vapor
street lights around the world.
I soon found out that orange-colored paper with black
ink is mostly black to that color of light and the camera always
had a tough time seeing enough contrast to render useful text.
OSU's school colors are orange and black.
A similar situation occurred if the paper was printed
with blue or purple ink. A common duplicating technique before
Xerox machines got cheap enough to be the preferred method was
something called Spirit Master or Ditto sheets. They produced
blue or purple letters in varying degrees of clarity which were
hard for even sighted people to read with their eyes.
Anyway, I remember that a lot of mail was very hard to
read and i mainly hoped to get a word or two out of the mess to
determine whether to take it home to my wife or just toss it
right then and there.
Martin
stuart young writes:
> Hi Martin.
> What would be the best OCR solution to have?.
=======================================
The Techno-Chat E-Mail forum is guaranteed malware, spyware, Trojan, virus and
worm-free
To modify your subscription options, please visit for forum's dedicated web
pages located at
http://mail.tft-bbs.co.uk/mailman/listinfo/techno-chat
You can find an archive of all messages posted to the Techno-Chat group at
either of the following websites:
http://mail.tft-bbs.co.uk/pipermail/techno-chat/index.html
Or:
<http://www.mail-archive.com/[email protected]>
you may also subscribe to this list via RSS. The feed is at:
<http://www.mail-archive.com/[email protected]/maillist.xml>
---------------------------------------
