> -----Original Message----- > From: ceo@xxxxxxxxx [mailto:ceo@xxxxxxxxx] > Sent: Thursday, December 18, 2008 10:06 AM > To: php-general@xxxxxxxxxxxxx > Subject: Re: Read/decode barcodes from an image > > The barcodes are on faxes and whatnot, with no predictable skew, > position, nor orientation. > > You've tried JOCR/GOCR, and they don't do very well. > > Here are your options: > 1) Shell out the money for that PaperPort OMNI or whatever it is > commercial OCR product. It *is* better than JOCR/GOCR, in my > experience. > [If you Google for OCR you'll find it, as it's the only/best contender, > and referenced everywhere.] > > 2) Roll your own with GD. With sufficient skill and time, you can > utilize very specific knowledge of your content to do even better than > the generalized commercial solution. I have done this myself for > standardized medical documents that were scanned in, getting a > percentage point or two better than 1) > > No matter what you do, OCR will never ever get you 100%. You WILL need > a human oversight process on the results that hand-checks everything, > or be prepared to accept a (small) failure rate. > > Set your [client's] expectations properly, or be doomed to frustration > [failure]. I had always thought that faxes were a big no-no for character recognition software? I know that we can't use it for one of our OCR solutions because it is just too... well, crappy. The image quality is nowhere near what it should be for a successful OCR pass. As for client expectations: hell yes--tell them not to use a fax machine in the digital age. Scan the document and e-mail it or something. Where I work, we've done away with 90% of any fax machine usage in order to have documents that can be used by our OCR solution. My 2c, // Todd -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php