You scanned a document, opened the PDF and tried to copy a paragraph, but nothing could be selected. Searching for a name finds nothing either. That is not a bug: a scanner or a phone camera saves a picture of the page. OCR, short for optical character recognition, reads that picture and turns the shapes into real letters so you can search, select and copy them.

Two common jobs

• You want to keep a PDF but make it searchable. Use the PDF OCR tool. The result still looks like your scan, but now the text is selectable and searchable.
• You have a photo or screenshot and only want the text. Use Image to Text, which turns a picture into text you can copy. It handles English and Hindi out of the box.

Steps

1. Upload the scanned PDF, or the image, to the matching tool.
2. Wait for recognition to finish. Long documents take longer, because every page is analysed.
3. Download the searchable PDF, or copy the extracted text.
4. Proofread. OCR is very good on clean print, but it is never perfect.

Get a better result before you run OCR

• Resolution: 300 dpi is the usual sweet spot for printed text. Below about 150 dpi, small letters blur together.
• Straight pages: a crooked photo makes lines tilt. Place the page flat and keep the phone parallel to it. You can rotate pages with the Rotate PDF tool first.
• Good light, no shadows: a shadow from your hand across the page is read as part of the picture.
• High contrast: black text on white paper works best. Faded stamps or coloured backgrounds lower accuracy.
• One language per page where possible: mixed scripts are harder to recognise than a single one.

What to check after OCR

Read through the numbers first. Look-alike characters are the classic errors: the letter O and the digit 0, the letter l and the digit 1, or 5 and S. Compare names, dates, amounts and reference numbers against the original. Tables and handwriting are the hardest cases; handwriting in particular is often not recognised reliably.

OCR and file size

A searchable PDF is as large as the scan plus a hidden text layer. If it is too heavy to upload, compress it afterwards, as explained in our guide on reducing a PDF to a size limit.

Frequently asked questions

How do I know if my PDF needs OCR?

Try to select text or search for a word. If nothing is highlighted or found, the pages are pictures and OCR is needed.

Does OCR work on handwriting?

Only partly. It is designed for printed text. Neat block capitals have a better chance than cursive writing, but always verify the output.

Which languages are supported?

English and Hindi are available out of the box. Other languages depend on the language packs installed, so check the tool page.

Will OCR change how my document looks?

For a searchable PDF the page image stays the same; the recognised text is added invisibly behind it.