[PLUG] Scanning and OCR in Linux

[John Jason Jordan]

On Sun, 9 Apr 2006 19:37:05 -0700
Aaron Burt <aaron at bavariati.org> dijo:

> On Sat, Apr 08, 2006 at 09:43:45PM -0700, John Jason Jordan wrote:
> > In the meantime, I guess I have to rely on Windows to do OCR. 
> 
> Shame, that.  The state of voice-recognition is pretty poor, too.
> Reading the Oregonian's "Ooh! Shiny! Windows Vista!" article, that was
> the one feature that wasn't already available on Linux.
> 
> Wine has a SANE-TWAIN interface, so Winders apps can use scanners in
> Linux.  I wonder if that could be made to work with an OCR app?  Might
> be a fun project for some Linux Clinic or something.

I later found the following site:

http://www.simpleocr.com/Demo/

If you upload a scan from XSane as a black and white TIFF at about 300 dpi it does a pretty good job on the bare text. Unlike the sophisticated OCR utilities available on Windows, however, it blows away pictures and even tables. Can't even handle tabs or columns. It also has no "learning" abilities. But it's better than retyping. And I used it for the 35 pages I needed to do, so I was able to avoid having to use Windows.

Pehaps a web-based interface is the future for OCR on Linux? Maybe it could be made a feature of OpenOffice.org? That way, while we're hitting Bill Gates in his cash cow, MS Office, we can hit Abbyy (FineReader) and Nuance (Omnipage) at the same time.