Thursday, May 24, 2007

New strategy for OCR

Ben Maurer is proposing a new way to 'OCR' old books. Words the computer can't OCR will be turned into CAPTCHAs.

A possible problem is that OCR programs often can't distinguish a word from a smudge, decorative gylph. Books often contain occasional Greek letters, nonsense, etc. So if the CAPTCHA implementation isn't good, Ben's strategy will add to the 19 person-years already wasted every day filling out annoying CAPTCHAs.

Still, it is a good idea and worth considering.

No comments: