How to Copy Text From an Image or Scanned PDF (OCR)
Text in a screenshot, a photo of a sign or a scanned contract looks like text but is really a picture. Optical character recognition (OCR) turns it back into characters you can copy, search and edit.
How to do it
- Open Image to Text (OCR).
- Drop one or more images, or a PDF.
- Choose the language of the text.
- Wait for the result, correct any mistakes in the box, then copy or download it as .txt.
The first run downloads the recognition engine (Tesseract) and the language model — a few megabytes — and then caches them. After that it starts instantly and even works offline. Your files never leave the device.
PDFs with real text
Many PDFs already contain text (exported from Word, for instance). The tool detects those pages and copies their text exactly, without OCR. Only scanned pages are read as images.
Getting accurate results
- Photograph the page straight on, in even light, without shadows.
- Fill the frame with the text; small text in a big photo reads badly.
- Keep "Improve the image first" on — it boosts contrast and enlarges small images.
- Pay attention to the confidence score. Below about 70%, check every line.
Limits
OCR is built for printed or typed text. Handwriting, very blurry photos, tables and multi-column layouts produce errors, and vertical Japanese or Chinese text is less accurate than horizontal. Supported languages include English, Portuguese, Spanish, German, Japanese, Korean and Chinese.
Related: turn PDF pages into images with PDF to JPG, or photos into a PDF with Images to PDF.