What's good in the way of OCR for scanned Word docs?

[cjolley]

We need to OCR some scanned paper into Word documents occasionally.
We have the scanner. (Cannon DR-3080C)
Windows XP.
I was looking about OmniPage but wanted to get some experienced advice if possible.
This is for a Government Office so accuracy is very important.
TIA
chuck

[Nowhere]

Recognita, from my experience, is very good (practically the only program which handles ¹, ê, ó, œ, ³, Ÿ, æ, ñ, ¿ (don't even try to pronounce this ) very good, although you probably don't care much about that )

[az]

There is not one single OCR program that doesn't make mistakes. You will always have to proof-read carefully

AZ

[RedRed]

I am with AZ here... in a previous life, however, the imaging solution we recommended was Omnipage...... BUT its not perfect...

(I does have strong linking into library/document retreval systems software however (which is good for the future)) - I would recommend that when you are doing OCR, you attach/embed the ORIGINAL picture to the document - that way - when(if) it gets to court with a suing case, you are in the clear - the original is instantly available...... (Been there, done that! )

RedRed

[cjolley]

Well, this is not for mass doc storage, we do that with our own home grown software.
It allows you to scan and catalog and store docs about as fast as you can cram them into the Cannon DR-3080C (which, by the way, I highly recommend)

The OCR software I am looking for is just for casual office use, for letters & stuff.
Definitely would be proof read.
I'll give Recognita a look.
As for the special characters, you'd be amazed at the weird stuff we get.
thanks
chuck

[az]

In the msot recent review I've seen, all three major OCRs scored quite near each other, IIRC - if you want, I'll try and dig it out.

AZ

[cjolley]

Well, now I'm looking at TextBridge.
It's cheap and simple.
Looks like it may have the same OCR engine as OmniPage.
OmniPage's ScanSoft seems to own Caere/Recognita now and use it's engine.

Who knows?
chuck

[UtwigMU]

IMO Scanning is still done best from image editor since it gives you more control.

Fire up Photochop (or other image editor)
Scan as 300 dpi grayscale (B&W photo)

1) Rotate canvas to straighten text.

2) Adjust levels (or brightness/contrast) so that the paper is white. (If you have a lot of pages, you can batch-automate this step, since all papers will scan at same settings and require same adjustment.)

3) Save as .tiff

Recognita (came with my Umax Astra 2200 U+S) is good for Word and properly scans Slovenian alphabet.

Open image, and ocr, paste or save as rtf.

Run over in Word with spellchecker.

[LvR]

Textbridge Pro 9

Calibrates itself according to the scanner being used and as long as the docs you wanna use is of good print quality, recognition is close to 100%

[Kurt]

-Abbyy FineReader 6.0 Professional

-OmniPage Pro 11.0 (12.0 out yet?)

I found Abbyy to be slightly better and to give you a faster way to transform your scans in proper documents. OmniPage is the easiest to find, but it's not cheap. I put the rest in the same pack, Textbridge used to be great years ago, now I can't say anymore as I stopped using it in '98 because it was sub-par.