How to Fix a Scanned (Image-Only) PDF
Standards this affects
- PDF/UA — ISO 14289-1, clause 7.1 (real, extractable text)
- WCAG 2.1 — 1.4.5 Images of Text (AA), 1.1.1 Non-text Content (A)
- Section 508 — 502 / WCAG 2.0 AA by reference
What this means
A scanned or image-only PDF contains a photo/scan of each page rather than real, selectable text. There is no text layer for a screen reader to read, no way to search or select text, and no structure.
Why it matters
To assistive technology, an image-only PDF is a blank page with one big picture. It's completely inaccessible until the text is recovered with OCR (optical character recognition) and the document is tagged. This is one of the most severe failures because no content is available at all.
How the checker flags it
- You can't select or search any text in the document.
- The checker reports no text content, or flags the page as an image.
- File size is large and pages look like photos/scans.
How to fix it
- 1
Run OCR to recover the text
In Acrobat Pro, use Scan & OCR → Recognize Text → In This File. This adds a real text layer behind the scanned image. Choose the correct language for best accuracy.
- 2
Proofread the OCR output
OCR makes mistakes, especially on poor scans, tables, and unusual fonts. Use Acrobat's "Recognized Text" review to correct errors, since screen readers will read exactly what OCR produced.
- 3
Tag the document
After OCR there is text but usually no structure. Auto-tag the document, then fix reading order, headings, and alt text as you would for any untagged PDF.
- 4
Prefer re-creating from the source
If you have the original digital file, re-exporting a clean, tagged PDF from it will always beat OCR on a scan. OCR is the fallback for paper-only originals.
Check your PDF for this issue
Upload your PDF to see whether it has this problem — and every other PDF/UA and WCAG 2.1 AA issue. Free, instant, no signup required.
Frequently asked questions
- Isn't OCR enough to make a scan accessible?
- OCR is necessary but not sufficient. It recovers the text, but the document still needs tagging, reading order, headings, and alt text to be genuinely accessible and PDF/UA compliant.