I took a look at MyScript, but the program does not accept PDF format.

I have a few documents (PDF) that are scans of hand writing.

I need a program that can take the PDF scan it, use OCR and convert it to text.

Anything like this exist?


Not really. Handwriting recognition is a hard problem and the failure to crack this was one of the key reasons that the Apple Newton failed. Palm had to make a pseudo-script called 'Graffiti' in order to recognize it, and that was done one letter at a time on a special pad, rather than written as a script.

You could try an OCR program - it might work, but at best you will probably have to spend quite a bit of time fixing up the errors made by the program.

An alternative might be to try something like rentacoder.com and see if you can pay someone to enter the text by hand. There are quite a few people in developing countries who work through this, so it might be possible to get it done quite cheaply. Alternatively, there might also be someone who has a Mechanical Turk application that does this.

  • 1
    Also don't forget that it is much easier to recognize handwriting when you actually enter it, where you have information about the strokes and such, than it is to make sense out of handwriting from scanned images.
    – Joey
    Jul 21 '09 at 11:10
  • Regarding Newton and Graffiti, I use Calligrapher on my WinMo phone and have found it very accurate without having to change my handwriting. Jul 21 '09 at 11:24
  • I guess you haven't seen my handwriting ... ;-} Jul 21 '09 at 12:17
  • Graffiti and the like get to use info about timing and stylus pressure which are not preserved when working from a scanned document. Jul 21 '09 at 14:01

If the documents aren't terribly long, you're better off doing this manually. Text recognition software is only so good for typed words, let alone handwritten words.

If you haven't written the text yet, you could use certain pen-based software solutions like ritePen. That's not the case here, though.


The type software you are looking for is called ICR (Intellgent Character Recognition) not so much OCR. However, it really only is very effective at printed text and mostly for those forms where you fill in boxes with one letter per box.

Like everything else, it ain't so hot for script.

Not the answer you're looking for? Browse other questions tagged or ask your own question.