How to make a scanned PDF searchable without uploading it
A scanned PDF is a stack of pictures, so Find and copy have nothing to work with. Text recognition (OCR) reads the words and writes them in as an invisible layer under each page. Free programs do it on your own computer, so the scan never has to leave it.
15 October 2026 · 6 min read
Why Find does nothing in a scan
A scanner produces pictures. A PDF made from them has a picture on each page and no characters at all, so Find, selecting and copying have nothing to work with. Some scanners’ software recognises the text as it saves and some does not, which is why two scans of the same letter can behave differently.
To tell which you have, try to select one word. If a word highlights, the page has text. If the whole page highlights as one block, or nothing does, it is a picture only.
What text recognition adds to the file
OCR (optical character recognition) reads the words in each picture and writes them into the page as text that is not drawn, each word placed over the picture of itself. The page looks exactly as it did. Find now lands on the right spot, a selection follows the lines, and copied text pastes as words.
It makes mistakes, more on poor scans than on good ones. A recognised page is fine for finding one letter in a folder of them. A figure you are going to rely on, such as an amount or an account number, is worth checking against the picture.
Free ways to do it on your own computer
- OCRmyPDF is free, open source and runs from the command line on Windows, macOS and Linux.
ocrmypdf scan.pdf searchable.pdfadds the text layer,-lchooses the languages, and--skip-textleaves pages that already have text alone. It uses Tesseract, the open-source engine most free OCR is built on. - NAPS2 is a free scanning program for Windows, macOS and Linux that recognises text as it scans, or when you import pictures or PDFs, and saves a searchable PDF.
- Adobe Acrobat Pro does it too, as part of a subscription.
Windows has two smaller helpers: the Snipping Tool’s text actions, and Text Extractor in Microsoft’s free PowerToys. Both copy the words from whatever is on screen, which is handy for a sentence or two, but neither makes the file searchable.
What the online route costs
Google Drive recognises the text of a PDF or picture opened with Google Docs, and many websites convert scans for free. Both mean the document is uploaded and read on somebody else’s servers, and kept there for as long as their terms say. For a contract, a medical letter or a passport, that is a lot to hand over to read a page.
Getting good results
- Scan at 300 dots per inch. Lower loses small print; higher makes larger files and rarely reads better.
- Keep the page straight and evenly lit. Most tools straighten a slightly crooked page, but a page curling into the spine of a thick book reads worse at its edge.
- Choose the document’s language. A page read with the model of another script comes out as nonsense, and one read in a related language loses its accents and special letters. A page in two languages wants both chosen.
- Keep the original scan. Recognition can be run again with a better tool later; paper that has gone cannot be scanned again.
Our own product, disclosed
Where our own software comes into this
We make Rutba Office, a free desktop suite for Windows whose PDF app has Recognise Text. It reads scanned pages on your computer, with Tesseract 5 built in and the models for English, Urdu, Arabic, Hindi, Chinese and Japanese in the download, so it works with the network off. The text layer is added after the file’s original bytes, which stay as they were, Undo takes it away, and pages that already have text are left alone.
On real English scans it gets about 2.3 characters in a hundred wrong; our documentation publishes the measurement for each language, and Urdu set in Nastaliq is still marked beta. Since version 1.51 it can also find a scanned invoice’s details and a form’s tick boxes, and turn the boxes into real check boxes.
Related
How to redact a PDF so the words are really gone
A black box drawn over words hides them on screen and leaves them in the file, where anybody can copy them out. Real redaction takes the words out of the page and out of everywhere else the PDF keeps them, and then you check the result by searching it.
How to turn a pile of receipts into a spreadsheet
Decide the columns first, keep each receipt beside its row, and check every row by its own sums: the items add up to the subtotal, and the subtotal and tax to the total. Reading the figures by machine saves typing, and the sums are what tell you whether to trust them.
How to fill in and sign a PDF without printing it
If the PDF has form fields, any good reader fills them in; if it has none, you type on top of the page. A signature placed as a picture is what most forms want. A digital signature is something else: a certificate proving who signed and that the file has not changed since.
Ten apps, one free download
No account, no telemetry, and it works with the network switched off.