How to extract text from a scanned PDF with OCR

You open a scanned PDF — a contract, an old invoice, a page from a book — and try to copy a line of text. Nothing selects. You search for a word and get no results. That's because a scanned PDF isn't really text at all: it's a picture of text. To get the words back, you need OCR.

Why you can't select text in a scan

When a document is scanned or photographed, each page becomes an image. Your eyes read the letters, but to the computer it's just a grid of pixels — there are no characters stored anywhere. So copying, searching, and screen readers all fail. A "born-digital" PDF (exported from Word, say) keeps the real text; a scan does not.

What OCR does

OCR — Optical Character Recognition — looks at the image, recognises the shapes as letters and words, and reconstructs the actual text. Good OCR can then do one of two useful things:

• Give you the plain text to copy and paste elsewhere.
• Make the PDF searchable by adding an invisible text layer on top of the scanned image — the page still looks identical, but now you can select and search it.

How to OCR a scanned PDF without uploading it

Scanned documents are frequently the sensitive kind — IDs, tax forms, medical records, contracts — so uploading them to an online OCR service is exactly what you don't want. Modern browsers can run OCR locally, on your device.

KeepItLocally's OCR PDF tool does the recognition in your browser, with no server involved:

• Open OCR PDF and add your scanned file.
• Let it process each page — recognition runs on your machine, so timing depends on your device and the number of pages.
• Get your result — searchable PDF and/or extracted plain text — and download it.

Getting the best OCR results

• Higher-resolution scans read far better than blurry phone photos. Aim for a clear, well-lit, straight page.
• Straighten skewed pages before OCR — crooked text confuses recognition.
• Clean, printed text is the most accurate; handwriting and decorative fonts are much harder.
• Proofread the output. OCR is very good but not perfect — always check names, numbers, and dates.

After OCR

Once your scan is real text, the rest of your toolkit opens up: convert it with PDF to Text, shrink an image-heavy scan using these compression tips, or combine several processed pages with Merge PDF — all without anything leaving your device.