Improving OCR performance

Question

I've sent a number of PDF documents to OneNote. The quality of OCR is poor - for example in a list of 60 company names only five have been correctly converted to text. 

Are there any tweaks to improve the performance of OneNote? 

I'm running Windows 7 and OneNote 2010. The PDFs have been created by other applications, but I have some tools for manipulating them if it would help improve OCR.

Answer 1

One Note 2007 was brilliant at OCR; so was Microsoft Document Imaging 2007. Now One Note 2010 OCR is very poor indeed.

Bear in mind that Send to One Note 2007 did not work at all on 64 bit computers. we had to buy 2010 to get that. The OCR performance is miles behind the 2007 versions now, whether 64 or 32 bit.

We need a complete re-issue of this part of Office 2010.

Answer 2

I'm having the same problem. We also used to have microsoft document imaging, and for ocr, ON2010 is a big step back.

It's not just company names and such, also for simple full text ON2010 performs really poor. It misses e's for c, spaces are wrong, brackets, i and j, t and l, ... I just did the procedure stated above, and the result is the same, although the files used are high enough resolution for ocr.

Answer 3

If you print the PDF to XPS, how many of the company names are found?  And if you right click the printout, copy it, delete it and then paste it back (a lot to do, I know), do all the names get recognized?

Answer 4

Ben,

I assume Onenote uses a thesaurus and some knowledge abouts semantics to interpret the result of OCR.

Is that correct ?

This would explain why company names are significantly more difficult to recognize.

Bernd

Answer 5

The cleaner the original the better the OCR.

Could you have Acrobat OCR them first before sending to OneNote?

---

-B-

http://www.officeforlawyers.com | http://www.onenote-tips.com

Author: The Lawyer's Guide to Microsoft Outlook