[sane-devel] XSane and Tesseract

[Jeffrey Ratcliffe]

Hi Mike,

On 4 August 2010 13:10, Mike CALDER <mikecalder at optusnet.com.au> wrote:
> DFRDB2.txt is the result that I have cut from the scan of the WHOLE letter
> using gscan2pdf and tesseract as the chosen ENGINE and you can see it is not
> as good as result.txt.

I suspect that your scan settings were not the same. The resolution
makes a difference - aim for 300 or 400dpi. If you pass tesseract an
image that does not have a depth of 1, then it thresholds it
internally. In my experience, either scanning it black and white, or
thresholding it afterwards (e.g. gscan2pdf uses imagemagick internall
to do this) produces better results.

> I may be wrong but even though there is a PREVIEW tab I do not seem to be
> able to do a preview scan and the select a PART for a partial scan.

gscan2pdf does not ATM have a preview option. I didn't need one for
ADF scans. The preview option you are seeing is probably an option
offered by the backend - which it probably just another way of asking
for a high speed, low resolution scan. Using the libsane-perl frontend
(this can be changed in Edit/Preferences), gscan2pdf just passes all
the options that the scanner backend offers.

Regards

Jeff