Many PDFs of old papers are only an image of text. Other papers have been converted with optical character recognition (OCR). By OCR'ed, I mean the text is electronically typeset, and so there are no artifacts from the scanner and good support for zooming in, copying-and-pasting, etc. In a properly OCR'ed document, the text layer is on top of the image level, so the reader can read the text comfortably; otherwise, the text will look ugly, as if the document has not been OCR'ed at all.

I find documents of the latter group to be very annoying. Is there any free service to fix such documents? Or other options to make reading such documents easier?

Example link: https://link.springer.com/article/10.1007%252FBF00117714

By OCR'ed and ugly, I mean the following image. Please click on the image to see the enlarged text: enter image description here

  • 7
    Before downvoting - consider this may be a barrier to people with poor vision or who rely on screenreaders, etc. – Azor Ahai -- he him Jul 29 at 16:39
  • 2
    There was great confusion about OP's goal here. I made my best guess and cleaned up; Hamidah, you can edit if necessary. – cag51 Jul 31 at 22:47
  • 1
    @cag51 The paper in my original post was OCRed but ugly. The image layer was on top of the text layer, so the text was searchable but super annoying to read. – Hamideh Aug 1 at 13:51
  • 3
    Yeah, sorry, I have no idea what distinction you are making. – cag51 Aug 1 at 17:59
  • 2
    I think you are using "OCR'ed" in a way that might not be what we are understanding. It sounds like you want old papers in a "modern" format that is properly typeset. OCR does not and has never done that. Do you want a service that updates old papers? Or do you only want an image converted to copyable text? The on top/below distinction is still not clear to me. – Azor Ahai -- he him Aug 3 at 15:38

An easy way that often works: Go to https://scholar.google.com and search for the paper and then click "All X versions".

I not only got an open access version (your link is paywalled), but it even was OCRed (the original image lies over the OCRed text, but you can copy & paste as usual): paper link

Another option would be to OCR it yourself. There are very good (and often quite expensive) commercial products, but you can try to use pdftk and tesseract/gocr on the command line as well. You'll probably also find easier to use frontends for the open source tools.

If you're looking for a nicely formatted document, you'll out of luck unless the author's (or someone authorized by the authors) published a newer version with modern formatting.
For equations no automated process will help because even when it mostly works someone needs to check the result afterward, as even small errors in recognizing parts of the equation are a huge problem.

| improve this answer | |

If by "ocr pdf files" you mean applying Optical character recognition so that they become searchable, you can either:

  • Look for a software to do it yourself. I personally use OCRFeeder with correct results, you can find more recommendations e.g. on writing.sxe or softwarecs.sxe.
  • Find a service on-line that runs one of those tools and do it for you. There is some discussion about this in graphicdesign.sxe

If you mean "typing it in a new document", then you'll have to find a human person to do it, as it cannot be automatized (as far as I know).

What specific software or on-line tool to use is a) Outside of this particular stackexchange website, b) likely to be treated as a "shopping question" and therefore closed.

| improve this answer | |
  • 1
    The full Adobe Acrobat (DC?) does a pretty good job of OCR. A few years ago I got an old copy of a 100 page report from 1962 ("typeset" on a typewriter), ran it through a scanner to make a PDF, then used Acrobat to OCR it. There were a few oddities, mainly from equations partly hand written, but it did a great job overall. – Jon Custer Jul 29 at 17:01
  • 1
    The paper linked in the question is OCRed and searchable. – user151413 Jul 29 at 18:08
  • 1
    @user151413 - good catch. So, using full Acrobat one goes and formats the text into another font (which will likely move stuff around, but...) that is more readable. – Jon Custer Jul 29 at 19:24
  • @JonCuster Indeed. Should work well for papers like the linked one, but less well for a math paper. – user151413 Jul 29 at 19:28

Your Answer

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

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