Re: [orca-list] inftyreader

[Michael Whapples <mwhapples aim com>]

As far as I know there is no Linux alternative for OCR on mathematical 
documents. Whether any of the general OCR systems for linux would be 
willing to try and spend time on adding maths support I don't know (I 
would guess not, as although inftyreader is said to work reasonably well 
it still has its moments where it can't do it). Below is a list of some 
of the OCR systems for linux, you may wish to ask the authors of those 
whether they would be willing to look at math support.

Cuneiform http://launchpad.net/cuneiform-linux I feel cuneiform gives 
the best results on general documents for me. The things I don't like 
about it include, its unable to correct page orientation automatically 
and sometimes if I put the blank side of the page on the scanner by 
mistake then cuneiform will crash very ungracefully and dump you back at 
the bash prompt.
Tesseract http://code.google.com/p/tesseract-ocr I haven't had quite the 
success people seem to be claiming with this and may be that's why I go 
for cuneiform. If the accuracy people report is correct then it seems to 
be reasonable. Also it has been opensource longer (I think) and other 
projects do make use of it more than cuneiform (eg. gscan2pdf).
Ocrad http://www.gnu.org/software/ocrad/ocrad.html This is the OCR 
software I used before I found cuneiform. Its enough to read things, but 
sometimes you get unusual output. I chose it over GOCR (mentioned below) 
for no significant reason, I think it was that it had less options and 
so seemed easier to use.
GOCR http://jocr.sourceforge.net (that's right jocr is the sourceforge 
name, I believe this difference is due to a conflict with another 
sourceforge project name). This is a very widely used opensource OCR 
system (supported by gscan2pdf, and I am sure plenty of other packages). 
Plenty of options to tweak, possibly too many for someone not sure on 
what to modify.

Any way that's a brief list of opensource OCR for linux. As OCR is not 
really an issue relating to orca, it may be seen as off topic and so may 
be in future other lists may be more relevant (eg. the NFB blindmath may 
have been a good place for asking about linux alternatives for 
inftyreader, as I assume its really the maths part of inftyreader you 
are interested in).

Michael Whapples
On -10/01/37 20:59, Timothy Taves wrote:
> I have been looking for a way to convert mathematical text (in scanned 
> and pdf format) to something I can read.  Windows has a program called 
> inftyreader which can uses ocr to convert documents to LaTeX code, 
> which can be understood by ear.  Is there an equivalent on program on 
> Linux?
>
> Thanks, Tim