Re: [Bug-ocrad] OCR without loading a file

[JS]

Antonio Diaz Diaz wrote:
> J.S. wrote:
> > I want to use OCR for recognition of text within images that are 
> > loaded from another software. The problem is that every OCR software 
> > I found
> > (including  Ocrad) seem to _force_ you to specify a filename etc.
>
> Ocrad does not force you to specify a filename. It can act as a filter.

Ok, so you can pipe a file in it then...

> > I thought maybe I had missed something, so here would be my twofold 
> > question:
> > 1) Is there a wait to use libjocr to process an image directly (e.g. 
> > an unsigned char*)
>
> I don't know. Libjocr is not part of Ocrad.

Oh sorry, i meant ocrad. Its just I posted a similar mail on the jocr 
mailing list and obviously forgot to replace part of it ;)

> > 2) If not: what would you suggest me to do? I'd be ready to 
> > contribute some code but a bit of guidance would help a lot.
>
> Try to execute ocrad through the `popen' function.

I actually started working on a fix using mkfifo().

Thanks for your help,

> Regards,
> Antonio.


--
++ js