dhare%integral@Sun.COM (Dwight Hare) (03/01/89)
I've recently purchased the OCR package Omnipage. Previously I had used Read-it. Omnipage is about twice the price of Read-it and is not trainable as Read-it is. The ease of use and accuracy of Omnipage is much better than Read-it in my experience. The major problems I had with Read-it was that the lines and characters were isolated before any character recognition was done. I found it impossible to find a contrast/darkness setting which would allow Read-it to separate characters accurately, without mistakenly separating some single characters in two. Omnipage obviously has algorithms to help separate the characters as well as recognize them. It also uses context to distinquish a "one" from a lowercase "l" and to distinquish a "zero" from an uppercase "O". I have only used Omnipage on magazine articles which use standard fonts. I don't know how flexible its character recognition algorithms are. A word of warning to those of you using or contemplating OCR packages: no matter how many obviously copyrighted news articles you've seen on the net, if you post something copyrighted, you'll get grief. Dwight Hare PS. I have no association with the developers of Omnipage or Read-it. The opinions expressed are my own.