Skip to content
PdfPress.online
How-to guides

How to OCR a Scanned PDF and Make It Searchable for Free

A scanned PDF is just pictures of pages. Run OCR to make it searchable, prep your scans for better accuracy and check what comes out.

PdfPress EditorialPublished Updated 6 min read
Share:
Illustration for How to OCR a Scanned PDF and Make It Searchable for Free
On this page

You scanned a 40-page lease, and now you need to find the clause about notice periods. Ctrl+F finds nothing, because to your computer every page is a photograph. A PDF OCR converter fixes this by recognising the letters in those images and adding them to the file as real, searchable text.

This guide covers running OCR on a scan with PdfPress, preparing pages so recognition is more accurate, and checking what comes out. Recognition happens in your browser, so the scans never get uploaded.

Does your PDF actually need OCR?

Check first, because plenty of PDFs that look scanned already have text. Open the file and try to drag-select a single word. If just that word highlights, there's a text layer. If nothing happens, or the whole page lights up as one block, it's an image.

Searching is another quick test: look for a word you can clearly see on the page. No match usually means no text. For a more thorough check, drop the file into PDF to Text; an empty or garbled result confirms it. Our article on scanned vs searchable PDFs explains the difference in more depth.

How to run PDF OCR on a scanned file

  1. Open OCR PDF.
  2. Drag in your scanned PDF or pick it from your device.
  3. Choose the document's language if the option appears. Getting this right makes a noticeable difference, especially for accented characters.
  4. Start recognition. Each page is analysed in your browser, so long documents take a while.
  5. Download the searchable PDF.
  6. Open it and search for a few words you can see, just to confirm the text layer is there.

Tip: Got a single photo or screenshot instead of a PDF? OCR Image pulls the text straight out of the picture.

How can you make your scans more accurate?

OCR engines are good, but they can't read what isn't clearly there. Five minutes of preparation often saves an hour of correcting.

Problem Effect on OCR Fix
Pages sideways or upside down Text may not be recognised at all Rotate PDF before OCR
Blank backs from double-sided scanning Wasted processing time Remove Blank Pages
Separate photos of each page Needs to be one document first Combine them into a single PDF
Very low resolution Letters blur together Rescan at around 300 dpi if you can
Shadows or skew from a phone camera Lines and characters misread Rescan flat, in even light, filling the frame

Phone scans deserve a special word. A page photographed on a kitchen table at an angle, with your hand's shadow across the bottom third, will produce noticeably worse text than the same page on a flatbed. If you're using a phone, lay the page on a dark, plain surface, shoot straight down and let the scanning app crop to the edges.

Note: Shrinking image resolution or converting to black and white makes files smaller, but doing it before OCR can hurt accuracy. Recognise the text first, then compress the result.

What you get back

The output is what's often called a "sandwich" PDF. The original scanned image stays on every page, so the document looks exactly as it did. Behind it sits an invisible text layer, with each recognised word placed where it appears on the page. When you search, the match highlights in the right spot; when you select, you're selecting that hidden text.

Archives and document management systems tend to prefer this format, since it keeps the scan as a faithful record while adding the convenience of text. Screen readers can use the text layer too. If you're curious about the mechanics, our explainer on how OCR works goes deeper.

Checking the results

No OCR is perfect, so verify whatever you'll rely on. Search for the names, reference numbers and dates that matter most. Copy a paragraph into a text editor and skim it; errors jump out much faster as plain text than they do hidden behind an image.

Some confusions come up again and again: zero and capital O, the digit 1 and lowercase l, 5 and S, and "rn" read as "m" (so "modern" becomes "modem"). Tables and multi-column layouts are the hardest to get in the right reading order. Headers, footers and rubber stamps are often in small or decorative type and come out less reliably than body text.

When one page comes out badly, the fix is almost always in the input. Rescan it straighter or crop away the dark border, run OCR on just that page, then swap it into the document.

Handwriting and mixed forms

OCR engines are trained mostly on print. Neat block capitals sometimes get recognised; cursive rarely does. On a form that mixes printed labels with handwritten answers, expect the labels to become searchable and the handwriting to be skipped or mangled. Worth knowing before you count on searching a stack of handwritten notes.

Warning: The recognised text sits hidden behind the image. If you cover something on a scanned page by drawing a black box over it, the text underneath may still be searchable and copyable. Use proper redaction instead, as explained in PDF redaction vs whiteout.

What a text layer makes possible

Once your scan is searchable, a lot opens up. You can convert it to Word to edit the wording, though complex layouts will need some tidying. Tables can be pulled out into a spreadsheet. Long documents become quick to search for a single figure. And because a text layer is one of the first requirements for screen reader support, OCR is often the starting point for making old scanned archives accessible.

Keeping scanned documents private

The things people scan tend to be the personal ones: signed contracts, medical letters, tax records, passports. Many OCR services want those uploaded to a server. On PdfPress, recognition runs inside your browser, so the scan stays on your device from start to finish. The trade-off is speed. Your own computer does the work, so a 100-page document can take several minutes on an older laptop. For most people that's a fair price for not handing their tax return to a stranger.

Written by PdfPress Editorial

Our editorial team writes and maintains practical guides on PDF workflows. Articles are reviewed for accuracy and updated when tools or best practices change. Spotted an error? Let us know.

Questions readers ask

Does OCR change how my scanned PDF looks?

With a searchable PDF output, no. The original page image stays visible and the recognised text is placed in an invisible layer behind it, so the page looks the same but can be searched and selected.

How accurate is OCR?

It depends heavily on the scan. Clean, straight, high-contrast printed text is usually recognised very well. Handwriting, faded copies, unusual fonts and low-resolution phone photos produce more errors, so always spot-check important numbers and names.

Is my document uploaded when I use OCR on PdfPress?

No. Text recognition runs in your browser on your own device. Processing speed depends on your device, so long documents can take a while on older computers.