How to Extract Text from Images, Screenshots & Scanned Docs (Free OCR Guide)

You've got a screenshot of a tweet, a photo of a whiteboard, a scanned contract, or a PDF of a handwritten form — and you need the text out of it. Optical Character Recognition (OCR) is the technology that makes this possible, and in 2026 it's fast, accurate, and free.

What is OCR and how does it work?

OCR (Optical Character Recognition) is the process of analyzing an image and identifying the text characters within it. Modern OCR systems use neural networks trained on millions of text samples to recognize characters even in challenging conditions: varying fonts, handwriting, poor lighting, skewed angles, and multiple languages simultaneously.

The output is text you can select, copy, edit, and search — extracted from an image that previously held it locked as pixels.

When is OCR actually useful?

More situations than most people realize:

• Scanned documents: physical contracts, invoices, and forms scanned as images become editable text
• Screenshots: grab text from a UI, a locked PDF, a social media post, or any screen where copy-paste doesn't work
• Business card photos: extract names, numbers, and emails without typing them manually
• Whiteboard photos: turn meeting notes and diagrams into text instantly
• Book or article photos: photograph a page and extract the text for notes or citation
• Handwritten notes: convert handwriting to typed text (accuracy depends on legibility)
• Foreign language text: extract text from images in other languages for translation
• Locked PDFs: some PDFs disable text selection; if you can take a screenshot, OCR can recover the text

Step-by-step: extract text from any image

Using ToolzPedia's Image to Text tool

Go to the Image to Text (OCR) tool.

1. Upload your image

Supported formats: JPG, PNG, WebP, GIF. Drop the image into the upload zone. The file is processed locally — nothing is sent to a server.

2. Select your language

The tool supports 17 languages including English, Arabic, French, German, Spanish, Portuguese, Urdu, Chinese (Simplified), Japanese, Korean, and more. Selecting the correct language improves accuracy significantly, especially for languages with non-Latin scripts.

3. Click Extract

Processing takes 2–10 seconds depending on image size and text density. The extracted text appears in the output panel.

4. Copy or download

Copy the text to clipboard with one click, or download it as a .txt file.

Getting the best OCR accuracy

Image quality is everything. OCR accuracy is directly proportional to image resolution and contrast. Best practices:

For photos of documents:

• Minimum 300 DPI for printed text; 400+ DPI for small fonts
• Shoot in good lighting — avoid shadows across the text
• Hold the phone parallel to the document to avoid keystone distortion
• Use your phone's document scanning mode (available in iOS Camera and Google PhotoScan) which auto-corrects perspective

For screenshots:

• Take full-resolution screenshots — don't resize before uploading
• High-contrast screenshots (dark text on white background) achieve near-100% accuracy

For handwriting:

• Print handwriting extracts better than cursive
• Consistent letter size and spacing helps significantly
• Dark pen on white paper, good lighting, no shadows

What OCR struggles with

Understanding limitations helps you work around them:

• Stylized fonts: highly decorative or handwritten-style fonts have lower accuracy
• Low contrast: light text on white, or dark text on dark backgrounds
• Rotated text: text at 45-degree angles, vertical text in narrow columns
• Small text in images: thumbnails with embedded text that's only a few pixels tall
• Complex layouts: tables with merged cells, two-column layouts with images
• Handwriting in mixed styles: cursive combined with print in the same line

When OCR produces errors in these situations, the output usually still gives you 80–90% of the text, making manual cleanup much faster than transcribing from scratch.

OCR for 17 languages: how it works

The tool uses Tesseract.js, a JavaScript port of the industry-standard open-source OCR engine maintained by Google. Tesseract includes trained models for 100+ languages, with particularly strong results for:

• Latin-script languages: English, Spanish, French, German, Italian, Portuguese
• Arabic script: Arabic, Urdu, Farsi (right-to-left text is handled correctly)
• CJK characters: Chinese (Simplified and Traditional), Japanese, Korean

For English text on a clean image, accuracy typically exceeds 99%. For less common languages or challenging images, expect 90–97%.

Practical workflows that use OCR

The "locked PDF" workaround

Some PDFs are configured to disable text selection and copying. If you can open and view the PDF, you can screenshot individual pages and run OCR on the screenshots to recover the text.

Business card digitization

Photograph a business card on a flat white surface. Upload to Image to Text. Copy the extracted name, number, and email directly into your contacts app. Faster than typing, and more reliable.

Whiteboard-to-notes

After a meeting, photograph the whiteboard immediately. Run OCR on the photo. Paste into your notes app. You now have a searchable, editable record of the session without transcribing.

Legal document extraction

Scanned contracts are often stored as image PDFs with no selectable text. OCR lets you extract clauses, dates, and names for reference in a document without manually retyping pages.

Free vs. paid OCR: what are you giving up?

Cloud OCR services like Google Document AI, AWS Textract, and ABBYY FineReader charge per page and offer advantages for:

• Very high volume processing (thousands of pages per day)
• Complex table and form extraction with structured output
• API access for automated workflows
• Highest accuracy on degraded documents

For individual use — extracting text from a few images, screenshots, or scanned pages — browser-based free OCR is completely sufficient. The accuracy difference between Tesseract and enterprise OCR is most pronounced on degraded scans and complex layouts; for clean modern documents, it's negligible.