On this page
How to OCR a PDF Online for Free: Extract Text From Scanned Documents
To OCR a PDF online for free, upload a scanned PDF to an OCR tool, let it recognize the page images, review the extracted text, and download the result. SelfPDF’s OCR Scanner runs recognition in your browser with Tesseract.js and exports plain text; it is useful for making image-based documents searchable in a text workflow, but it does not recreate a perfectly formatted, searchable PDF layer.
Published 2026-10-17 · Updated 2026-10-17
What is OCR?
OCR, or optical character recognition, turns letters in an image into machine-readable text. A normal text PDF contains a text layer that you can select and search. A scanned PDF may contain only page images, so copying, searching, indexing, and accessibility tools cannot see the words until OCR analyzes those images.
OCR is an interpretation, not a pixel-perfect transcription. Recognition quality depends on scan resolution, contrast, skew, language, typography, page layout, and whether the source is printed or handwritten.
How to OCR a PDF online with SelfPDF
Step 1
Open the OCR Scanner
Go to the [SelfPDF OCR Scanner](/tools/ocr). The tool accepts one PDF at a time and is intended for scanned-document text extraction.
Step 2
Select a scanned PDF
Drop a PDF into the file area or choose it from your device. Use a copy if the original is important.
Step 3
Start text extraction
Select the extract action. SelfPDF renders PDF pages as images and recognizes them with Tesseract.js in the browser.
Step 4
Review each page
Check headings, names, dates, numbers, punctuation, columns, and page breaks. OCR can insert or omit characters even when the output looks plausible.
Step 5
Download the text
Download the extracted `.txt` file or copy the result into your editor. If you need Markdown from an ordinary text PDF, try [PDF to Markdown](/tools/pdf-to-markdown); for scanned pages, OCR first.
Can OCR make a PDF searchable?
OCR makes the words searchable only when the result is placed into a searchable text layer or an indexed text file. SelfPDF’s current OCR Scanner exports extracted plain text rather than writing a new invisible text layer back into the original PDF. You can search the downloaded text, edit it, or use it as a starting point for a new document.
If you specifically need a searchable PDF that preserves the original page appearance, use a PDF OCR workflow that supports creating a text layer and verify the output in a PDF viewer. Do not assume that downloading OCR text has changed the original PDF.
Scanned PDF to text: what works best?
- Printed pages with sharp, high-contrast text usually produce the most reliable results.
- 300 DPI scans are a useful target; very small text may need a higher-resolution source.
- Straight pages with clear margins are easier to recognize than skewed or tightly cropped pages.
- Typed Latin text is generally easier than handwriting, decorative fonts, mixed scripts, or dense tables.
- If the PDF already has selectable text, use PDF to Markdown, PDF to Text, or another text extraction workflow instead of OCR.
Common OCR errors and how to fix them
| Problem | What you may see | What to do |
|---|---|---|
| Similar characters | O/0, I/1/l, S/5, rn/m | Search for likely substitutions and compare against the scan. |
| Layout loss | Columns become one continuous paragraph | Read in page order and rebuild headings, columns, and tables manually. |
| Punctuation errors | Missing commas, quotes, or decimal points | Proofread names, dates, measurements, and numbers against the image. |
| Noise and shadows | Random characters or broken words | Use a cleaner scan, crop dark borders, or increase contrast before OCR. |
| Handwriting | Incomplete or nonsensical words | Treat output as a draft; handwriting recognition is much less predictable. |
| Language mismatch | Accents and words are misread | Use an OCR engine and language model that support the document language. |
How to improve OCR results
- Scan the original at a readable resolution and avoid photographing pages at an angle.
- Use even lighting and remove shadows, glare, folds, and dark borders.
- Deskew and crop pages before recognition when the source is tilted or surrounded by background.
- Process a short sample first. Compare the result before spending time on a long document.
- Review every critical value manually. OCR should not be the sole verification step for contracts, invoices, medical information, or legal records.
- For complex tables, diagrams, multi-column layouts, and forms, expect to rebuild some structure after extraction.
OCR versus PDF text extraction
PDF text extraction reads an existing text layer. OCR reads page pixels when that layer is missing or unusable. A text-based PDF can often be converted directly with PDF to Markdown, while a scanned PDF may need OCR Scanner first. Running OCR on an already selectable PDF can introduce errors instead of improving it.
For a long document, a practical workflow is to test a few representative pages, choose OCR only for image-only pages, then proofread the output and retain the original scan.
Privacy and sensitive documents
SelfPDF’s OCR Scanner performs the recognition work in the browser using Tesseract.js, so the PDF is not uploaded to a SelfPDF OCR server in the normal client-side workflow. Browser processing still exposes the file to the page in memory, and your browser, extensions, device, and deployment matter. Do not treat this as a universal privacy guarantee for every SelfPDF feature.
AI tools such as summarization, translation, and Chat with PDF have a different data path and may send the text needed for the requested task to an external AI service. Use the specific tool notice and your organization’s policy before processing confidential content.
What to do after OCR
- Search and correct the extracted text before quoting it.
- Use PDF to Markdown to structure text-based content for notes or AI workflows.
- Use PDF to Word when you need an editable office document and the conversion supports your file.
- Keep the original scan so you can resolve ambiguous characters later.
- Name the OCR output clearly and record the source pages if the text will be used in research or compliance work.
Frequently asked questions
How can I OCR a PDF online for free?
Open SelfPDF’s OCR Scanner, select a scanned PDF, run text extraction, review the result, and download the extracted text. The current tool exports plain text rather than embedding a searchable layer into the original PDF.
How do I extract text from a scanned PDF?
Use OCR because a scanned PDF usually contains images rather than selectable text. OCR recognizes the page images and produces machine-readable text that you should proofread.
Can OCR make a PDF searchable?
Yes, when OCR text is added as a searchable layer or indexed separately. SelfPDF’s current OCR Scanner exports text for searching and editing; it does not automatically rewrite the source PDF with an invisible text layer.
Does OCR work on handwriting?
It may recognize some handwriting, but results are substantially less predictable than printed text. Treat handwriting OCR as a draft and verify it against the page image.
What is the best scan quality for OCR?
Clear, straight, high-contrast scans work best. Around 300 DPI is a useful target for printed pages, though small text and difficult originals may need more resolution.
Is SelfPDF OCR private?
SelfPDF’s OCR Scanner performs recognition in the browser with Tesseract.js in the normal client-side workflow. Review the specific tool behavior and your own browser or organizational environment before processing sensitive files.
Try it now
Extract text from a scanned PDF with a practical OCR workflow.
Use a focused PDF workflow that keeps the next step simple and your documents organized.
Try SelfPDF OCR