How to OCR a scanned PDF so you can search and copy the text

A scanned PDF is a stack of photos. Without OCR (optical character recognition) you cannot search it, copy from it, or quote it. OCR reads the shapes of letters in the images and adds an invisible, real text layer on top.

Step 1: check what you actually have

Open the file and try to select a word. If the cursor grabs text, your PDF already has a text layer and you do not need OCR. If it draws a selection box around a photo, it is a scan and OCR is the way forward.

Step 2: run the OCR tool and pick the language

Open the OCR tool, drop the scan, and choose the document's language. Language data matters: running English models over a Japanese scan produces confident nonsense. The recognition runs on your device; a scanned contract never needs to be uploaded anywhere.

Step 3: verify against the original

OCR is a probability engine, not a typist. Spot-check names, numbers, and amounts against the page image, especially in documents that matter. Straight scans of clean printed text usually land near-perfect; low-contrast receipts and handwritten notes will not.

What helps accuracy

• Higher scan resolution: 300 DPI is the sweet spot; under 150 DPI accuracy drops fast.
• Straight pages: rotate crooked scans before OCR.
• One language per pass: bilingual documents do better with two passes than one mixed model.

After OCR: what you get

The output is a searchable PDF you can search with any reader, or plain text you can paste anywhere. Searching for 'invoice 1042' across hundreds of pages only works if the text layer exists — that is the entire point.