OCR PDF

A PDF from a scanner or a phone app holds one picture per page and nothing else. To your computer, the words on it are just patterns of dark pixels, which is why search finds nothing and copying selects the whole page as an image. OCR, optical character recognition, changes that by working out which letters those patterns are and where each word sits on the page.

The recognized words are then written into the PDF as real text, positioned exactly under the words in the picture and drawn invisibly. PDF viewers treat that text like any other: search jumps to it, selection highlights it and copy pastes it. Screen readers can read it aloud, which makes scanned documents accessible to people who cannot see them.

Because the text layer sits exactly under the picture, a searchable PDF is also the safest way to make a scan usable without losing anything. Nothing about the page changes for a reader who only looks at it, while anyone who needs to search, quote or index the document gets the words.

Searchable PDFs are what archives, law offices and accounting departments want from scanned paperwork, because a document nobody can search is a document nobody finds. The PDF format itself has been an ISO standard since 2008 as ISO 32000-1, and its archival profile, PDF/A, since 2005 as ISO 19005-1. Both define how text and images live in the same file, so a searchable scan made today will still open and search the same way in many years.

The list above shows what this page accepts and what it can produce. Scanned PDFs are the usual input, but photos and image scans work too: each one comes back as a searchable PDF page, which is a quick way to turn a phone photo of a contract into a document that can be filed and searched.

If an archive or court system asks specifically for PDF/A, run the searchable PDF through PDF to PDF/A afterward. That step adds the extra rules PDF/A requires, such as embedded fonts and color information, without touching the text layer the OCR added.

Many records offices also ask that scanned files be searchable before they are filed, since full-text search is how they are retrieved later. Checking one or two words with your viewer's search box before you submit confirms the layer is there.

OCR reads what the scanner captured, so the scan settings matter. Printed text recognizes reliably from scans at around 300 dpi; much lower resolutions make small print ambiguous, and much higher ones only make the file bigger. Black-and-white or grayscale scans are fine for text documents and keep files small, while color is worth keeping only where color carries meaning.

Straighten pages and Auto rotate are on by default and fix the most common problems with fed and phone scans: pages slightly askew and pages upside down. Enhance contrast under Advanced settings helps with faded photocopies, thermal receipts and gray paper. For documents in more than one language, add the second language in the same panel.

Long documents are fine. Each file can be up to 95 MB, and up to 20 PDFs can go through in one batch with the same language and settings. For a scan with a different language in part of the document, run those pages separately with a page range and the right language for each part.

A searchable PDF keeps the document's appearance. When you need the words in another form, change Output: plain text gives you only the words, Markdown keeps headings and lists in a light format for notes apps, and Word rebuilds the layout as an editable document. The image to Word page is preset for that last case.

For the general tool with all outputs, use the Image to Text tool. Your files are removed when processing ends and the results 30 minutes later, and you do not need an account.