A scanned PDF is a picture of a page, so a redaction has to cover pixels, and when text recognition has been run it also has to delete the text layer. The method has four parts: find the text, cover it, clean the file, and test the saved copy.
Why a scanned PDF needs its own method
Adobe explains that a paper document scanned to PDF contains only image data and no searchable text. Optical character recognition (OCR) converts the image text into selectable, searchable text, and Acrobat creates a searchable text layer in the PDF. A scanned file therefore comes in two kinds. An image-only scan holds pixels and nothing else. A recognised scan holds the same pixels and a layer of text that the viewer does not show.
Redaction has to work on both. Covering the pixels hides the words from the eye and leaves the text layer if there is one, and deleting the text layer leaves the pixels if the image was never touched. A redaction that treats only one of the two leaves the other in the file.
How to redact a scanned PDF step by step
- Keep the original and work on a copy.
- Find out which kind of file you have. Try to select a word with the cursor, or search for a name. If nothing is selected, the file is image-only; if the word is found, there is a text layer.
- Mark what has to go on every page: names, addresses, identification numbers, account numbers and the handwriting, signatures and stamps that no search will find. Look at the back of double-sided pages.
- Apply the redaction with a tool that deletes the content under the box. In Acrobat Pro the steps are All tools, Redact a PDF, Redact text and images, drag over the text or image, then Apply.
- In the dialog that opens, switch on the option to sanitise and remove hidden information, and save the sanitised file under a new name.
- Run the four checks of the table below on the saved file.
- If a check fails, repeat the work starting again from the original.
| Check | How | A pass looks like |
|---|---|---|
| Search | Search the saved file for each surname and for the first digits of each number, including the account and tax numbers | No result |
| Select all | Select all text, copy it and paste it into a plain text editor | None of the removed words appears |
| Zoom | Zoom to 400 per cent on the edge of each box | No pixel of a letter shows around the box |
| Second viewer | Open the file in a different PDF viewer, and open the properties | Same result, and no author or title that names the client |
Each check answers a different failure. The search finds a text layer that survived, the select-all finds text that the search missed, the zoom finds a box that sits slightly off the word it covers, and the second viewer finds what one program hides and another shows. A scan can pass the first two and fail the zoom when the box is a few pixels too small.
Adobe advises reviewing the document after text recognition to make sure the text is accurate and complete, correcting errors manually or running the recognition again with adjusted settings.
What OCR does in the redaction workflow
OCR makes a scan searchable. In Acrobat the flow is All tools, Scan & OCR, In this file, then the page range and the language, then Recognize Text. After that you can search for a surname and mark every hit at once, which is the one way recognition helps the redaction. It has a limit that shapes the checks. If the recognition reads "Schneider" as "Schnelder", the search misses the word and nothing is marked, so the page still has to be looked at. This is why step 3 asks you to mark by eye on every page, and why the first row of the table is a search while the fourth is a second pair of eyes on the result.
What goes wrong with scanned documents
The report of the National Security Agency on safe publication names three common mistakes: covering text with black rectangles, covering parts of an image with black rectangles, and leaving metadata and document properties in the file. It says the first two are effective for printed pages and generally not for computer documents distributed over networks. A scan is both a picture and, once recognised, a text document, so it can fall into the first two mistakes at once.
The National Archives of the United Kingdom shows what a failed box looks like in practice. In the judgment Bromcom Computers Plc v United Learning Trust [2022] EWHC 3262 (TCC), black boxes stand for redacted text and a reader can still select the text underneath, although the text underneath was only x characters. The guidance asks courts to replace the text with [REDACTED], because boxes do not prevent the text from being accessed.
A third failure is the leftover metadata. Adobe lists metadata, comments and hidden layers among the content that sanitising removes, and a scan that passes through several tools can collect them.
How Redactic redacts a scanned PDF
Redactic reads the text inside scans and photos with OCR, so the names and numbers it finds are replaced with placeholders without a separate recognition step. In Review and edit you see every item on the page and draw a box over anything the recognition missed, such as a signature, a stamp or a handwritten note. A box drawn in Review and edit removes the content and the image pixels under it. The option Black out all images covers the embedded images of a PDF or Word file and leaves a scanned page to the text redaction, so on a scan the box you draw does that work. You can upload a PDF, a photo, a TIFF or a ZIP of up to 200 files, and documents are never stored and never used to train AI models. The AI-safe documents page shows the flow, and the free trial takes your own document. If the file has hidden properties, read hidden metadata in Word, PDF and Excel, and for a document type that may arrive as a scan, how to redact a payslip.
Frequently asked questions
Why does a search find a word under a black box on a scanned PDF?
Because the file holds a text layer next to the picture. Adobe explains that a scan contains only image data until text recognition adds a searchable text layer, so a PDF that was recognised earlier carries the words as text. A shape drawn over the page leaves that layer in place. Use a tool that deletes the content under the box, and run the checks in the table above on the saved file.
Do I need to check every page after OCR?
Yes. Adobe advises reviewing the document after text recognition to make sure the text is accurate and complete. A name that the recognition misreads is not found by a search, so it is not marked for redaction, and only a look at the page shows it.
Does a black box on a scan delete the picture of the text?
Only if the tool removes the pixels. A rectangle placed over an image changes how the page looks and leaves what is underneath. The National Security Agency report on safe redaction says that covering parts of an image with black rectangles is generally not effective for computer documents. In Redactic, a box drawn in Review and edit removes the content and the image pixels under it.
Can I redact a photo or a signature on a scanned page?
Yes, by drawing a box over it in Review and edit. The option Black out all images replaces the embedded images of a PDF or Word file, such as a photo or a signature, and it leaves a scanned page to the text redaction, because the scan is the page. See the AI-safe documents page for the flow.
Sources
- Adobe, Recognize text in scanned documents (Acrobat), accessed 2026-11-09.
- Adobe, Redact sensitive content in PDFs in Acrobat Pro, accessed 2026-11-09.
- Adobe, Sanitize PDFs and remove hidden content in Acrobat Pro, accessed 2026-11-09.
- National Security Agency, Redacting with Confidence, report I333-015R-2005, 13 December 2005, accessed 2026-11-09.
- The National Archives, Redactions, guidance for the Find Case Law service, accessed 2026-11-09.