How to Extract Text from an Image: A Simple OCR Guide
A photo of a page, a screenshot of a notice, or a scanned document all hold text that you cannot edit until it is converted. This guide explains what OCR is in plain words, how to take or pick an image that OCR can read well, the steps to run OCR in a browser tool, and how to fix the errors that appear most often.
What OCR Is in Plain Words
OCR stands for optical character recognition. It is software that studies the shapes of letters in an image and turns those shapes into real text. Before OCR, a photo of a page is only a picture: you can look at it, but you cannot search it, copy a sentence from it, or edit a word in it. After OCR, the same words become ordinary text that you can paste into a document, translate, or read aloud with a screen reader.
OCR is useful in everyday situations. Students copy notes from a whiteboard photo, shopkeepers digitise a printed rate list, and office workers turn a scanned letter into a document they can edit. The quality of the result depends far more on the quality of the image than on which tool you use, which is why the next section matters most.
Picking a Clear Image
OCR reads dark, sharp letters on a light background. Anything that hides the letters, bends them, or blurs them will lower accuracy. Use the table below to check an image before you run OCR.
| Problem in the image | Why it hurts OCR | Simple fix |
|---|---|---|
| Blur or camera shake | Letter edges blend together | Hold the phone steady, rest it on a book, and retake the photo |
| Shadow or glare | Parts of the text turn too dark or too bright | Move near a window in daylight and tilt the page away from direct light |
| Tilted or curved page | Lines of text bend and letters change shape | Photograph straight from above with the whole page inside the frame |
| Tiny text or low resolution | Small letters lose the detail OCR needs | Move closer or scan at a higher resolution so the text fills the frame |
| Busy background | Patterns behind the text look like letters | Crop the image to the text area only before running OCR |
If the original is a photo, crop away desk edges, fingers, and unrelated objects. If the original is a scan, a straight scan at good resolution beats a tilted phone photo every time. For a document that is already a PDF, the OCR PDF tool can read the text directly without a photo step.
Steps to Run OCR in a Browser Tool
- Prepare the image using the checks above, and crop it to the text you need.
- Open a browser OCR tool and upload the image or PDF file.
- Pick the language of the text if the tool asks for it, because the wrong language setting causes extra errors.
- Run the extraction and wait for the full result before copying anything.
- Copy the text into a document and proofread it against the original image, especially numbers, names, and codes.
- Save the corrected version, and keep the original image in case you need to check a doubtful word later.
Most browser tools need no installation and no account. If an image is very large, a smaller file may process faster; the image compressor can reduce file size, but keep the text sharp, because heavy compression that blurs letters will hurt OCR accuracy.
Image Readiness Checker
Tick each box that is true for your image, then press the button to see how ready it is for OCR.
Tick the boxes above and press the button to see your result.
Fixing Common OCR Errors
Even a good image produces a few errors, and the same mistakes repeat. Check these first during proofreading.
| Common error | Likely cause | What to do |
|---|---|---|
| O read as 0, or l read as 1 | Similar letter shapes in the font | Check every number, code, and amount against the original |
| rn read as m | Two narrow letters joined in a small font | Read names and uncommon words slowly during proofreading |
| Missing or jumbled words | Faint print, handwriting, or a crease in the page | Retake the photo with better light and fill the gaps by hand |
| Table columns merged into one block | Plain text has no columns | Check row by row and rebuild the table by hand |
| Wrong words in a heading or logo | Decorative fonts and large stylised letters | Retype headings and logos manually instead of trusting OCR |
Handwriting deserves a special warning. Printed and typed text converts well, but joined up handwriting, unusual letter shapes, and crossed out words often produce errors or missing words. For handwritten notes, treat OCR output as a starting draft and correct it by hand.
Best File Types for OCR
PNG suits screenshots and sharp text because it keeps edges clean. A high quality JPG suits phone photos and is easy to share. PDF suits multi page documents and scanned files, and a scanned PDF can go straight into an OCR PDF tool. Avoid heavily compressed images where the letters already look fuzzy, because OCR cannot restore detail that compression removed.
A Note on Privacy
Some browser tools process the file on your own device and state this clearly on the page. Others upload the file to a server for processing. Before uploading a photo that shows personal details, such as an identity card, a bank statement, or a private letter, read what the tool says about storage and deletion. When in doubt, crop the image to only the text you need, or use a tool that states clearly that files stay on your device. Never upload a document you would not want stored elsewhere.
Good OCR is mostly good preparation: a straight, well lit, tightly cropped image, followed by a careful proofread of numbers and names. Follow that routine and most extractions need only small corrections.
Frequently asked questions
What is OCR and how does it turn a photo into editable text?
OCR stands for optical character recognition. It is software that looks at the shapes of letters in a photo, screenshot, or scanned page and converts those shapes into real text that you can copy, edit, and search. The result is ordinary text, so you can paste it into a document, translate it, or read it aloud with a screen reader.
Why does OCR mix up letters like O and 0, or l and 1?
Some letters and digits look almost identical in many fonts, especially O and zero, lowercase L and digit one, and the pair r and n which together can look like m. Low resolution, blur, or a small font size makes the confusion worse. A clear, high resolution image and a quick manual check of numbers, names, and codes after OCR will catch most of these errors.
What kind of image gives the best OCR result?
A straight, well lit photo of dark text on a plain light background gives the best result. Hold the camera steady, keep the whole page inside the frame, and avoid shadows, glare, and tilted angles. A PNG or a high quality JPG works well for photos, while a scanned PDF works well for documents. Crop away desk edges and fingers before running OCR.
Can OCR read handwriting accurately?
OCR works best on printed or typed text. Neat handwriting may be partly recognised, but joined up writing, unusual letter shapes, and crossed out words often produce errors or missing words. For handwritten notes, expect to correct the result by hand and treat OCR as a starting draft, not a finished copy.
How do I fix OCR errors in a table or a receipt?
Tables and receipts are hard for OCR because the columns and rows carry meaning that plain text loses. Run OCR on a straight, high contrast photo, then check the result row by row against the original, paying special attention to totals, dates, and item codes. Rebuild the table by hand in a document or spreadsheet once the text is correct, rather than trusting the raw OCR layout.