How to Extract Text from a Scanned PDF or Image (OCR)
A scanned document or a photo of a page is just an image to a computer — there's no text to select, search, or copy, even though a human can read it fine. OCR (optical character recognition) is what bridges that gap. Here's how to use it well, and where it hits real limits.
How OCR works, briefly
OCR Tool analyzes the shapes in your image or scanned PDF and matches them against known letterforms to reconstruct the actual text — the same underlying approach used by every mainstream OCR engine. The result is plain, searchable, copyable text extracted from what was previously just pixels.
Step by step
- Upload your scanned document or photo — clear, well-lit images with the text held flat and in focus produce the most accurate results.
- Run OCR and let it process the image.
- Review the extracted text against the original — OCR is very good but not perfect, and it's worth a quick pass, especially for numbers, names, and anything you'll rely on.
- Copy or download the result as a plain text file, or feed it into PDF to Word afterward if you need a formatted, editable document rather than raw text.
Getting accurate results
- Use high-contrast images — dark, clear text on a plain light background reads far better than faint or low-contrast text.
- Keep the page flat and the camera or scanner straight-on — a skewed or curved page distorts letterforms and increases misreads.
- Avoid busy or textured backgrounds behind the text, and make sure text is at least about 10pt-equivalent in size in the image.
What OCR can't do
Handwriting is the big one — OCR is built around recognizing consistent, printed letterforms, and handwriting varies too much between people (and even within one person's writing) for general OCR to read reliably. Heavily stylized fonts, very low-resolution images, and text at extreme angles also degrade accuracy significantly.
Frequently asked questions
Does OCR work on handwriting?
Not reliably. OCR is built to recognize printed text — regular fonts with consistent shapes. Handwriting varies too much from person to person for general-purpose OCR to read it accurately.
Why did some words come out wrong?
Low image resolution, skewed or rotated pages, and low contrast (light gray text, busy backgrounds) are the most common causes of misread characters. A cleaner, higher-resolution, well-lit scan fixes most accuracy issues.
What can I do with the extracted text?
Copy it directly, download it as a .txt file, or run the same document through PDF to Word afterward if you want the recognized text in an editable, formatted document instead of plain text.