Hopefully I am understandable and don't sound impractical when I ask the above question. To elaborate: I have some scanned documents and the font used in those documents(images) is FrenchScript MT Std Regular. I installed that font on my sys. Now I was wondering if it's possible to use the character map of that particular font and match each individual character in the image (the scanned image is just plain text, no images, no paragraphs or no formatting) with font characters and place the matching letters just above the character in the image.

Want to do it in MS word or any other text editors where after the characters are placed above the image, it can be deleted from the background and the document can be saved. It's another type of OCR I guess. How can it be done? Attaching the image to help you understand the scenario.

the source scanned doc

  • 1
    Sounds like fairly typical OCR. Have you searched for "trainable OCR"? – Andrew Leach Jan 21 '14 at 7:09
  • I am with Andrew, since this is printed it should be reasonably possible with a good OCR. What have you tried? Have you tested some of the programmes available out there? – benteh Jan 21 '14 at 10:43
  • @boblet I did try with some programs and being OCR programs they are doing their job but not very accurate in the output. They are mistaking the tails of letters for commas and periods and many more such mistakes. Cuz there are so many pages, I can't really afford to recheck each page and that would be really tedious. Do you think the Tesseract-OCR would be of any help here? How are they used? And plz don't forget the main question. – Dave Jan 21 '14 at 15:06
  • I am not an expert on this at all, but my professor used a programme with reasonable success some years ago; on old, handwritten documents both in latin and arabic. I will see if I can find it, I think it was Tesseract. As to your wishes of an overlay functionality - I have no idea if that can be automated (you could always do it "manually" with the image "behind" your actual text. – benteh Jan 21 '14 at 16:08
  • Since your Q is about an overlay functionality; I will just give some pointers to the first leg: how to get text from image: my professor used Tesseract, he trained it (you can train it on different fonts). This is possible by ICR (intelligent character recognition). See this FAQ: code.google.com/p/tesseract-ocr/wiki/TrainingTesseract3 Further; for you overlay-idea, my only suggestion is to do this after conversion, putting the image behind the text, and then quite easily visually check errors. – benteh Jan 21 '14 at 16:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.