How to extract text from an image with free OCR
Retyping text from a photo of a document, a screenshot or a scanned page wastes time and introduces mistakes. Optical character recognition, or OCR, reads the printed words in an image and turns them into text you can copy, edit and search.
Image to text runs OCR in your browser. The picture is read on your device and is not uploaded to PixnPDF, and there is no daily limit.
Start with a clear picture
OCR is only as good as the image. Photograph the page straight on, in even light, with the text filling most of the frame. Avoid shadows, glare and blur. If the image includes a lot of background, crop it first with Crop image.
Screenshots usually work very well because the text is already sharp.
Choose the right languages
Select the language of the printed text before you start. Image to text supports English, Arabic, Urdu, Persian, Hindi, Bengali, French, Spanish, German, simplified and traditional Chinese, Japanese, Korean, Turkish and Russian. You can combine up to five languages, which helps with documents that mix, for example, English and Arabic.
Each language pack downloads once from this website when you first select it. Your image is not part of that download.
Review and correct
After recognition, read through the text and correct any mistakes before you export. Pay extra attention to names, numbers, dates and amounts, because a single wrong character can matter.
Basic browser OCR is designed for printed text. Handwriting is much harder to recognise and results will be unreliable.
Choose an output
Export plain text to copy into an email or document, a basic Word file, Markdown, or JSON with word positions. For a document you want to keep as a picture but make searchable, export a searchable PDF, or use Image to searchable PDF directly.
If your source is already a PDF, OCR PDF does the same job page by page and keeps the result as a PDF.
Typical uses
Copy a paragraph from a photo of a book or a printed letter into an email. Pull the text out of a screenshot so you can search it later. Turn a scanned notice into editable text so you can translate or quote it. Extract the details from a printed receipt into a note.
For a whole multi-page document, scan it to a PDF first with Scan to PDF and then use OCR PDF, which keeps the pages together and can export a searchable PDF.
For text in tables, such as a printed price list or statement, PDF to Excel can be a better choice because it keeps rows and columns.
Improve accuracy
If results are poor, retake the photo closer and straighter, or increase the light. Choose only the languages that actually appear, because extra languages give the engine more ways to misread characters. For pages with several columns, crop one column at a time so the text is read in the right order.
Small text benefits from a higher-resolution image. If a screenshot is tiny, zoom in on the original before you capture it.
How it works
- Take or choose a clear, straight image and crop away the background.
- Open Image to text and add the image.
- Select the languages printed in it.
- Run recognition, then correct names, numbers and dates.
- Export as text, Word, Markdown, JSON or a searchable PDF.
Before you export Always proofread OCR output before using it for anything official. Recognition can confuse similar characters such as 0 and O, or 1 and l.
Image to text · OCR PDF · Image to searchable PDF · Crop image
Common questions
Is my image uploaded?
No. OCR runs in your browser and the image stays on your device.
Does it read Arabic and Urdu?
Yes, along with Persian, Hindi, Bengali and several other languages. Select them before recognition.
Can it read handwriting?
Browser OCR is built for printed text, so handwriting results are unreliable.
Is there a daily limit?
No. Image to text is free and unlimited.