Some other considerations for OCR, from Microsoft Office Online
(http://office.microsoft.com/en-us/assistance/HA011021601033.aspx):

Black and White from color page: This preset should give the best results
for standard documents such as magazine pages, letters, book pages. The
page
is scanned in color and the image is then converted to black and white
before OCR is performed on the image. The resulting image file size is
smaller than if you scanned and stored it as color.

Black and white: If you want to perform OCR only in black and white (due
to
time, memory, or processor power considerations), scanning at 300 dpi is
best. Using this preset with a resolution setting of 200 dpi is also
acceptable for Western languages (for Asian languages, 200 dpi is not
recommended.)

Color or Grayscale: In general, if you scan your images using either of
these presets (at 200 dpi) regardless of the language the document uses,
you
will get good results.

Font point size: Text in 10-12 point font sizes will give the best OCR
results. Text in font sizes smaller than this, especially in Asian
languages, are not as easily recognized and will have worse OCR results.
Text in larger than 72 point font may not be recognized since at the
larger
sizes the font may be recognized as a picture within the document rather
than as text.

Text color contrast with background: Documents with high contrast between
the text color and the background have the best OCR results. Light colored

text on a light background, and dark colored text on a dark background
will
most likely not be recognized. For this same reason, text over a picture
may
not give good OCR results.

Density of text: OCR results will be best for images that contain
continuous text in context. Sparsely spaced text may not give good OCR
results.


Taina Makinen
Vital Records Specialist
Canadian Tire Corporation
[log in to unmask]

List archives at http://lists.ufl.edu/archives/recmgmt-l.html
Contact [log in to unmask] for assistance