Bakki Kudva writes:
> teri wrote:
>
> In its current state OCR seems to work only in very controlled
> situations. eg. a warranty return card which you have control over and
> you force the user to write into boxes, or neatly typed paper documents
> which law firms use to do full text OCR and indexing etc. However free
> form OCR from various qualities of underlying images is error prone and
> the fix non-trivial. Even in production environments OCR finds use only
> in some apps. I am afraid that the average home user has such a wide
> variety of receipts and paperwork that you'd have to manually check OCR
> results anyhow. You wouldn't want $80.00 entered as $8000 from a poor
> original. Also some times the receipts don't have ALL the information.
> You'd have to deduce where you bought something based on your own memory
> since the cash register receipt may not have a merchant name or anything
> on it.
>
Also, there isn't any open-source OCR software available AFAIK, and it
would be most preferable (both for technical and philosophical
reasons) to base this system completely around open source, rather
than have a key component proprietary.
If you *are* looking for a challenging project to make your name,
open-source OCR is probably a good one :)
------------------------------------------------------------
Robert Merkel [EMAIL PROTECTED]
"We can build a better product than Linux" - Jim Allchin,
chief Windows Technologist, Microsoft Corporation.
------------------------------------------------------------
_______________________________________________
gnucash-devel mailing list
[EMAIL PROTECTED]
http://www.gnumatic.com/cgi-bin/mailman/listinfo/gnucash-devel
