Using Numeric Regions When You're Recognizing Numeric Text; Putting Pages In The Scanner Properly - Visioneer PROOCR100 User Manual

Tips for Getting the Best Results
Using Numeric Regions When You're Recognizing
Numeric Text

Putting Pages in the Scanner Properly

file:///C|/VisioneerDoc/html/08tips.htm (12 of 13) [1/20/2003 4:21:24 PM]
Even with good quality characters on good quality paper, Pro OCR will have
trouble locating and recognizing accurately if the type in the page image is skewed
(crooked). This can happen either because the text is crooked on the page or
because the page is scanned at an angle.
What's important is that the text image cannot be skewed more than 2° for Pro
OCR to have accurate recognition. The illustration to the left shows a page that has
2° skew.
If text is skewed (at an angle) on the page, both locating and recognizing may be
affected. When text on a page is badly skewed, Pro OCR may have trouble
correctly locating paragraph boundaries and recognizing the contents of these
paragraphs. There are two ways to fix this problem:
You can adjust the paper so that the text is scanned in straight, or is not
skewed more than 2° . If text is straight on the page, make sure that the
paper is put in the scanner straight.
You can select the Straighten Skewed Images processing option. When you
select Straighten Skewed Images, Pro OCR can automatically rotate the
image of the page up to 15 degrees in order to straighten the text on the
page.
Use numeric regions for text in your document that will only consist of numbers.
Pro OCR will make sure that all characters in the numeric region are recognized as
numbers and not mistaken for letters.
Make sure that you put pages in the scanner with the proper orientation—Portrait or
Landscape—and use the corresponding Get Page setting.