It depends on a number of things, including the quality of the original
physical document and thus the resulting scanned image; the size and to some
extent the font used; and the quality of the software being used. If you're
buying your OCR software at your local office supply store for $29 and
scanning photocopies of thermal faxes that use Gothic script, your accuracy
may suffer a bit. :) 

On the other hand, nice clean laser prints that use common fonts at 10+
point can easily hit 99% accuracy. Most of the vendors will claim 98-99%
accuracy for standard machine print. And if that's not accurate enough,
image cleanup, barcodes (1D or 2D), field validation (ZIP and SSN can only
have numeric values), data bridges that validate the OCRed text on the fly,
or database lookups can all increase the accuracy to approaching 100%. Some
solutions also use neural network-type learning technology to get better or
use multiple OCR engines in the background and compare the results to
determine a confidence interval. These are pretty common as you move towards
the higher end of capture solutions. 

Regards, 

Jesse Wilkins
[log in to unmask] 

List archives at http://lists.ufl.edu/archives/recmgmt-l.html
Contact [log in to unmask] for assistance
To unsubscribe from this list, click the below link. If not already present, place UNSUBSCRIBE RECMGMT-L or UNSUB RECMGMT-L in the body of the message.
mailto:[log in to unmask]