Training Ocr - ScanSoft OMNIPAGE PRO X FOR MACINTOSH Manual

t
To remove a zone template:
• Select a non-template setting in the Original Layout pop-up menu
on the OCR Toolbar.
OmniPage Pro will no longer place template zones on incoming
page images. This does not remove template zones from existing
zoned pages. Just delete or modify them or choose Discard Current
Zones and Find New Zones in the Zoning Instructions dialog box.

Training OCR

You can create a
consistently misrecognized. A training file is a set of up to 256 pre-
recognized character shapes each linked to an OCR solution.
OmniPage Pro compares the stored shapes with those encountered on
incoming documents.
OmniPage Pro X is a powerful, pre-trained OCR product. For
recognizing ordinary characters in everyday fonts, training files should
not be needed. Training is useful mainly for long documents (or a set
of documents) in which a few character shapes are being repeatedly
misrecognized in the same way. Training is not useful for poorly
formed characters unlikely to occur again in the document. For
instance, a character shape damaged by spots on the image is a poor
candidate for training. Do not attempt to create a training file for an
unsupported language or alphabet.
t
To create a training file:

Open an image file or scan a page that includes the characters you
want to train or use a page you have already recognized.
If you select a recognized page, its recognition results are deleted.
Accept the invitation that appears when you finish, to re-recognize
the page with the new training file.

Create or modify zones on the page image if you want to train
characters from only part of the page.

Select Train OCR as the option in the OCR pop-up menu.
to handle characters that are being
training file
Chapter 5
Training OCR 97