Re: tesseract: ocr that works
On Sat, Dec 27, 2008 at 5:59 AM, Dotan Cohen <dotancohen@gmail.com> wrote:
> 2008/12/21 Hugo Vanwoerkom <hvw59601@care2.com>:
>> [3] don't scan at less than 300 dpi
>
> And don't scan above 600 DPI!
>
> I forget which OCR I played with a few years ago, but 300 and 600 DPI
> yielded satisfactory results. 1200 DPI made things _worse_ not better,
> possibly because of noise. This was on Fedora, so maybe it was in fact
> tesseract.
Back when I first got access to the university scientific publication
network, I started to get hungry for an OCR tool to do bibliographies
and references, here's the result with tesseract:
http://heybryan.org/shots/2008-03-24-autoscholar-OCR-notgood.png
I should have recorded the specific commands, citation, and so on, so
this now squarely falls under 'anecdotal' instead of being useful to
anybody. Sorry. But clearly that's pretty terrible.
- Bryan
http://heybryan.org/
1 512 203 0507
Reply to: