On Wed, 5 Oct 2005 09:36:40 +0100 Kevanf1 wrote:

On 05/10/05, Doug McGarrett wrote:

...

Is there a really good OCR for SuSE? What's it called and how do you get it and run it? Will it work for a UMAX 2200 scsi scanner? Will it read and/or output .pdf files?

(I never could get the scanner to work in older versions of SuSE.) It works fine in various Windows versions, but I don't have a really good OCR for Windows. And I went out of my way to buy a scsi scanner, so it might work on Linux. No such luck.

I will shortly (I hope) have SuSE 10.0.

--doug

I may be wrong but surely it would be the version SANE that you are using not the version of SuSE as to whether your scanner would work. Or am I wrong?

I've had two different SCSI scanners working with SANE on my old RH 9.0 box. This was a really long time ago, so I don't remember how I did it, but I do remember that it took about five minutes-- literally. Try typing "sane" at a prompt. hth, ken

-- ============================================== I am only human, please forgive me if I make a mistake it is not deliberate. ============================================== Take care. Kevan Farmer

34 Hill Street Cheslyn Hay Staffordshire WS6 7HR

-- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com

-- A lot of us are working harder than we want, at things we don't like to do. Why? ...In order to afford the sort of existence we don't care to live. -- Bradford Angier

On 06-Oct-05 Charles philip Chan wrote:

On 4 Oct 2005, dmcgarrett@optonline.net wrote:

...

Is there a really good OCR for SuSE? What's it called and how do you get it and run it? Will it work for a UMAX 2200 scsi scanner? Will it read and/or output .pdf files?

Xsane and Kooka both support external command line OCR programs. Two common ones are ocrad and gocr- two are both include with SUSE.

Charles

Thanks for these pointers -- I hadn't met ocrad before. I just compiled the atest ocrad and gocr from source. It looks as though ocrad may be slightly the better of the two, in two respects. 1. It's more compact (compiles to 300K as opposed to 535K for gocr) and is quicker. 2. It seems to do marginally better recognition. I fed ocrad an A4 page of double-spaced 11-point Times Roman, scanned at 300dpi to a BPM file. There were errors in 6 places out of 2400 characters, so 1 in 400. Gocr 0.4 scored 12 errors on the same file. Also, orcrad picked up ligatures (e.g. "fi" ) correctly, while gocr failed on these. Interestingly, an old version of gocr (0.31 from 2001 which I already had) also scored 6 errors like ocrad on the same file (but still failed on the ligatures). A 1/400 error rate isn't bad. You're going to have to check any OCR results anyway. Running the output past ispell is a quick way of cleaning up. On the other hand, maybe gocr offers more flexibility. Best wishes to all, Ted. -------------------------------------------------------------------- E-Mail: (Ted Harding) Fax-to-email: +44 (0)870 094 0861 Date: 06-Oct-05 Time: 20:19:44 ------------------------------ XFMail ------------------------------

On 6 Oct 2005, thorpflyer@yahoo.com wrote:

How do I make ocrad work from xsane? I find when I try to configure this that ocrad complains:

bad magic number - not a pbm file.

Scan in lineart mode. Here is some info about pbm: http://netpbm.sourceforge.net/doc/pbm.html Charles -- die_if_kernel("Penguin instruction from Penguin mode??!?!", regs); linux-2.2.16/arch/sparc/kernel/traps.c

On 7 Oct 2005, thorpflyer@yahoo.com wrote:

Strange. Lineart mode, of itself, doesn't fix the behavior. However, if I scan in lineart mode, then save the file using the pbm extension and "determine file type by extension" then I can use ocrad on the resulting file manually. Good enough, I don't do this often, but I'm a little curious why I can't configure the menu to do it directly.

I usually just scan it in "Viewer" mode in XSane and invoke ocr from the "File" menu of the viewer. Charles -- The nice thing about Windows is - It does not just crash, it displays a dialog box and lets you press 'OK' first. (Arno Schaefer's .sig)

On Thu, Oct 06, 2005 at 06:58:57PM -0400, Charles philip Chan wrote: [...]

If you you really want accuracy, this one is even better:

http://freshmeat.net/projects/claraocr/

However, you need to train it.

I can second that - especially on bad scans (I had to deal e.g. with scans from typewriter on grey paper), ClaraOCR produced much better results than the other two. Unfortunately, it is quite unstable (so save often!) and development seemingly has all but stalled for the moment. Cheerio, Thomas