Recognize text (OCR)
A scanned page is only a picture. Tools ▸ Recognize text (OCR)… reads the words in it and adds them as an invisible text layer, so you can search and copy the text. The page looks exactly the same. In the app: Help ▸ Show me.
Steps
- Open the scanned PDF and choose Tools ▸ Recognize text (OCR)….
- Choose the pages: Current page, All pages, or Pages with a range such as
1-3, 7. - Leave Skip pages that already contain text ticked to save time, or clear it to read every chosen page.
- Click Recognize. A progress window shows "Recognizing page 2 of 10…" with a Cancel button.
- When it finishes, a message reports how many words were found on how many pages. Use Find and search to check.
The result is one undoable page operation. Save the document to keep it.
The scanned-document banner
When you open a document whose first three pages have no text and are mostly covered by images, a bar says "This looks like a scanned document." Click Recognize text to open the dialog with All pages selected, or Not now to dismiss it. It shows once per document per session, and not for read-only documents.
To turn it off, clear Offer to recognize text when a scanned document opens under Edit ▸ Preferences… ▸ Documents. See Preferences.
Limits
| Item | Detail |
|---|---|
| Language | English only |
| Internet | Not needed; the engine is built in |
| Resolution | Pages are read at 300 dpi, lower for very large sheets |
| Speed | Runs on your computer, page by page; long documents take a while |
| Memory | About 100 MB while it runs |
Note
Handwriting, low-quality scans and non-English text give poor results. Characters outside the basic Western set may appear as ? in the hidden layer.
Important
The command is unavailable for encrypted (read-only) documents. If the engine files are missing, "Text recognition is not available" appears; reinstall the application.
If nothing is found you see "No text was recognized". If the pages already have text you see "All the selected pages already contain text".