6 comments

[ 2.6 ms ] story [ 26.5 ms ] thread
Does it work? Last time I tried a Google OCR it didn't work at all (forgot the name, something with t).
You're probably thinking of Tesseract.

I've actually been doing some work with various OCR tools the past few days. Tesseract and Ocropus didn't work very well for me, but I don't think mine is the intended use. I have a need to identify images that contain text (which could be on top of a complex background) and most OCR applications are better suited for reading scanned documents. So far GOCR has produced the best results for me.

I found an article a couple weeks ago comparing a few OCR applications for scanning documents that someone might find interesting: http://www.linux.com/feature/138511

OCR quality is heavily dependent on font type, font size and contrast, but for many typed documents Ocropus and Tesseract do a pretty decent job.

In that article, tesseract seems to work the best. Maybe I should give it another try.
What languages/alphabets does this support?
I tried these out for converting image-only PDFs, but the installation was getting ever more complex and unstable. I made a time-is-money decision and purchased Vividata - *nix based command line utility - which is a shame as I would've liked to have got these working. Almost all the other products were Windows based - urrgh! If anyone has a different experience, I'd be pleased to hear from them...
I wonder if there's a market for an OCR web service?