How to redact a scanned PDF so the text is really gone

To redact a scanned PDF, read the pages with OCR so you can find the text, mark what has to go, then burn each box into the page image and drop any hidden text layer under it. KeptPDF does all three in your browser. Open the Redact tool, drop the scan in, and it reads the pages on your device. Your file is never uploaded.

Why a scan is harder to redact than a normal PDF

A normal PDF stores its words as text. A scanned PDF stores a photograph of the paper. Try it on yours: press Ctrl+F (Cmd+F on a Mac) and search for a word you can see on the page. If nothing is found, and you cannot drag-select a line, the page is a picture.

That causes two problems.

• Nothing tells you where the sensitive words are. A tool that looks for Social Security numbers or names in the text layer finds nothing, because there is no text layer. To find them, something has to read the picture first. That is OCR.
• A box drawn over a picture is not a redaction. In many editors the black rectangle is a separate object lying on top of the scan. Remove it and the page underneath is untouched. And if the scan was ever run through OCR, the file also carries an invisible text layer behind the picture, so the words you covered are still there to search and copy.

Step 1: find what to redact with OCR

OCR (optical character recognition) turns the picture of a page into text a program can search. In KeptPDF it is automatic. When you open a scanned PDF in the Redact tool, it notices there is no readable text and reads the pages for you, with a progress bar and a Cancel button. Then it runs its detectors over the result: names, ID numbers, account numbers, emails, phone numbers, addresses, dates and more, shown as boxes for you to review.

• It runs on your device. The reading engine loads from keptpdf.com the first time you use it, then the pages are read in your browser. Your file is not sent anywhere. Open your browser’s developer tools, watch the Network tab, and your document never appears in a request.
• It straightens crooked pages. A page scanned a few degrees off level is straightened before it is read, and a page fed in sideways or upside down is turned upright first. Without that, OCR reads sideways text as nonsense.
• It only reads the pages that need it. If your file mixes typed pages with scanned ones, the typed pages keep their real text and only the scans are read.
• It reads English. That is the only language available today.

OCR is a first pass, not a promise. A smudged number, a fax, a faded stamp or handwriting can be misread or missed. That is why every box is yours to confirm, and why you can also click a word or drag a box over anything it did not flag, such as a signature or a handwritten note.

Step 2: make the redaction real

Here is what happens when you apply. The OCR step gave KeptPDF a working copy of your scan with an invisible text layer on it, which is how it found the words. That layer is exactly what must not reach the file you send. So the redacted pages are not edited. They are redrawn: each page is rendered to a picture, the black boxes are painted into the pixels, and the page is saved as a new image. The original text layer, and the original pixels under each box, are not carried over.

By default KeptPDF does this to every page, which is the safest setting. If you would rather keep the real text on pages you did not touch, the arrow beside the download button has an Only redacted pages option. Pages with no box then keep whatever text layer they had.

The same step clears the document’s hidden details, such as title, author and keywords, so a name left in the file properties does not give you away.

Step 3: check that it worked

KeptPDF checks its own work. After it builds the redacted file, it re-opens that exact file and confirms the redacted pages hold no extractable text. You can run the same kind of check yourself in about a minute:

1. Search for a covered word. Press Ctrl+F (Cmd+F) in the redacted PDF and type a name or number you boxed. No match is what you want.
2. Select all and paste. Press Ctrl+A, Ctrl+C, and paste into a plain text editor. A flattened scan pastes nothing, apart from the small KeptPDF footer a free download carries and any text layer you chose to keep.
3. Look at every page, zoomed in. This is the check only you can do on a scan. A box can sit a little off, or OCR can miss a word, and the file will still pass a text search. Read the pages the way the person you are sending them to will.

For the long version, see how to check if a PDF redaction worked.

How to redact a scanned PDF in KeptPDF

1. Open the Redact tool and drop in your scan. It runs in your browser, with no install.
2. Wait while it reads the pages. You will see “Reading scanned pages” with a progress bar. Keep the tab open until it finishes.
3. Review the boxes. Confirm the ones that are right and remove the rest. Click a word, or drag a box, to cover anything it missed.
4. Click Apply, then Redact and download. The pages are redrawn with the boxes burned in, and the file is checked.
5. Download the redacted PDF. Run the checks above on it before you send it.

Want people to be able to search the result? Open the arrow beside the download button and turn on Keep text searchable. KeptPDF then reads the finished, already redacted pages with OCR and adds a fresh text layer. Because it is read from the redacted pixels, it cannot contain words that are no longer on the page.

What to watch for

• Bad scans give bad OCR. The blurrier the page, the more it misses. Review those pages by eye.
• Pages that stay unreadable. If OCR cannot read enough text, KeptPDF says so, and you redact those pages by hand.
• Pictures are pictures. A photo of a licence, a handwritten line or a stamp is not text. Draw a box over it yourself.
• A plain black box fails on normal PDFs too. To see why, read how to redact so the text is actually gone.