Turn a scan into a searchable PDF

A scan is a picture of a document. This reads the words off it so the file can be searched and copied from, and makes sure what comes out is a size that will send.

• Pages that already carry text are taken as they are, not re-read.
• Recognition runs on this device, so nothing is uploaded.
• Ends under a megabyte, which is what mail servers expect. A scan that was already small ends a little larger, because the words get added to it.

A page that can already be read is not re-read

A PDF downloaded from a portal already carries its own text, and running a recogniser over it makes the result worse rather than better: the words come off the page exactly as whoever produced the file wrote them, which is more accurate than any recogniser and free.

So each page is checked. Where the words can be harvested off the page and the page does not look like a photograph, the recogniser is not run on it at all, and the panel says the text was taken from the page rather than read off it. A document that is typeset all the way through can be made searchable with no recognition at all.

Two conditions, both load-bearing. The page must not look like a photograph, because a scan somebody already ran through OCR has a text layer too and its geometry is only as good as whatever wrote it. And the page must carry several lines, or a cover page with a logo and a title would be read one way and the rest of the document another.

A small scan comes out larger

Making a document searchable means writing the words into it, and words take room. A 35 KB scan came out at 81 KB here: a text layer was added and nothing was compressed, because there was nothing to gain. The run says so rather than reporting a reduction that did not happen.

The size step is a cap and not a promise to shrink. A file already under the limit is left as the better copy, because compressing it is destructive and would buy nothing.

Why size is part of the same job

The reason to make a scan searchable is almost always that it is about to be sent somewhere, and a scan is large. Doing both in one pass means the file that gets attached is the one that was worked on, rather than whichever of three similarly named copies was nearest.

The compression aims at the number and keeps the largest result that fits, and the size it reached is reported rather than assumed. Compression is destructive, so open the saved file before sending it, because only you can say whether the text is still readable. The Compress tool draws page one on screen for exactly that, before anything is saved.

Is there a page limit?

The limit for one run is shown on the panel. There is no daily cap and no account.

Will the text be selectable afterwards?

Recognition makes it searchable. Note that compression redraws pages as pictures, so keep the original if you need to copy from it.

Support and billing: support@gosmartpdf.com. Operated by Mrityunjay Kumar, India. Payments by Dodo Payments, our merchant of record.

Loading the editor…