OCR to use - Page 2

pepak · 05-26-2008, 01:29 PM

Quote:

Originally Posted by Nergal

For the inital question: I recommend to have a look at tesseract ocr - it is an opensource command line tool - with an amazing recognition rate (95-99.9 %, mostly at 98-99% for me).

I am afraid Tesseract is not for me. I need some additional languages, I'd like a better accuracy, and the most important, I need a reasonable layout detection - the software must, at least, be able to detect paragraphs and store each on one line. That alone is worth the price difference for me. Thanks for the suggestion, though.

Nergal · 05-26-2008, 03:17 PM

The paragraph detection is tricky with tesseract but (!) not complete hopeless, if the paragraphs are seperated by a blank line it might be detected and would be parsable as two linebreaks. Though if I understand you correctly, you do not want something to tinker with, but a solution that actually solves a task

).

I just made a test with a book page I scanned today (Caesar - Civil War, German) - 2124 Signs, only 2 signs which stood together were read falsely, scanned at 300 dpi.
So that is a rate of 99.91% (better than my typing

.

Recently I had the chance to see ReadIris (for free with an HP-All-in One Device) - its layout detection was really horrible - very distinct columns have been overlooked and a lot of simple misreadings.

I had some years ago AbbyyFineReader 8.0 Trial and must admit I was a bit disappointed at the automagical layout-detection, quite a lot of manual editing. Hopefully this works better by now.

IIRC they offer an educational discount ... - if I wasn't a Linux-addicted at that time I surely would have bought it because of its fantastic recognition rate, except for text written in italics.

a bt OT: Their language support was/is awesome on the other hand: scanned in a Russian article, parsed it through babelfish and got at least a vague idea what the author had written, whithout knowing much more than 'spassibo' of Russian language by myself

Good Luck!

Nergal

igorsk · 05-26-2008, 05:30 PM

Saw just today on Teleread: Cuneiform open-sourced their OCR:
http://www.cuneiform.ru/eng/index.html

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Ebook readers - should you OCR or not?	crackhammer	Calibre	13	09-06-2010 02:32 AM
OCR Software Help	kpfeifle	Workshop	5	03-01-2010 02:27 PM
Unutterably Silly Memorable OCR errors	Patricia	Lounge	4	02-16-2010 02:53 PM
OCR help needed	Nate the great	Workshop	7	09-21-2009 11:21 PM
What is an OCR Cradle?	JackieFrost	Which one should I buy?	4	05-21-2008 08:10 PM

05-26-2008, 03:17 PM	#17
Nergal eBuchReisender Posts: 41 Karma: 208 Join Date: May 2008 Location: Münster Device: Palm Tungsten-E, iLiad	The paragraph detection is tricky with tesseract but (!) not complete hopeless, if the paragraphs are seperated by a blank line it might be detected and would be parsable as two linebreaks. Though if I understand you correctly, you do not want something to tinker with, but a solution that actually solves a task ). I just made a test with a book page I scanned today (Caesar - Civil War, German) - 2124 Signs, only 2 signs which stood together were read falsely, scanned at 300 dpi. So that is a rate of 99.91% (better than my typing . Recently I had the chance to see ReadIris (for free with an HP-All-in One Device) - its layout detection was really horrible - very distinct columns have been overlooked and a lot of simple misreadings. I had some years ago AbbyyFineReader 8.0 Trial and must admit I was a bit disappointed at the automagical layout-detection, quite a lot of manual editing. Hopefully this works better by now. IIRC they offer an educational discount ... - if I wasn't a Linux-addicted at that time I surely would have bought it because of its fantastic recognition rate, except for text written in italics. a bt OT: Their language support was/is awesome on the other hand: scanned in a Russian article, parsed it through babelfish and got at least a vague idea what the author had written, whithout knowing much more than 'spassibo' of Russian language by myself Good Luck! Nergal

05-26-2008, 05:30 PM	#18
igorsk Wizard Posts: 3,442 Karma: 300001 Join Date: Sep 2006 Location: Belgium Device: PRS-500/505/700, Kindle, Cybook Gen3, Words Gear	Saw just today on Teleread: Cuneiform open-sourced their OCR: http://www.cuneiform.ru/eng/index.html