🚀 go-pugleaf

In article <92Oct22.185130.20262@acs.ucalgary.ca> kthompso@acs.ucalgary.ca  
(Kelly Thompson) writes:
> So why can't you throw away the data that isn't needed. On a 600
> dpi scan, couldn't every other pixel be ignored, or interpolated
> with the pixel beside it?

That's basically what I suggested in my follow up as well.

> While we're at it, is it difficult to
> make the software recognize, say 2 bit grey-scale, and just turn
> a dark pixel black, and a light grey pixel white.

No, it isn't, and it's crap that OCR Servant doesn't do it
already.  Count me in as complaining about that too! 

> Again, these files are often produced somewhere else. There's
> compression so they aren't too huge.

That doesn't really apply here.  The concern is over the
amount of image data presented to the OCR process, and
compression doesn't change that.  The OCR software needs
to look at the image as pixels and compression (in
general) doesn't change the number of pixels, only their
representation.


Removing the density and image depth limitations of OCR
Servant would make it a much more useful service.  That's
uncontestable.  It would require less knowledge on the
part of the user, make the interface smoother, and might
even increase functionality.  Very NeXT things to do.

These changes >are< possible.  That's certain in my mind,
because I have a hack front-end program that does these
things.  (PBMPLUS would do most of it.)

The question then is: Is HSD going to fix it? 

Allen B (Putting his mouth where his money is)