x
Skip to main content

OCR (Optical Character Recognition)

Tool ID: ocr-pdf

Make scanned PDFs searchable and editable by recognizing text in images. Stirling-PDF's OCR tool uses Tesseract OCR engine to extract text from image-based PDFs and convert them into searchable, selectable documents.

What is OCR?​

OCR (Optical Character Recognition) is technology that recognizes text within images and converts it into actual, selectable text. This makes scanned documents searchable, editable, and accessible.

When You Need OCR​

  • Scanned paper documents (no text layer)
  • Photos of documents or whiteboards
  • Screenshots of text
  • Image-only PDFs
  • PDFs where you can't select or search text

How to Use OCR​

  1. Upload Your PDF - Select a scanned or image-based PDF
  2. Select Language(s) - Choose the language(s) in your document
  3. Configure Options - Adjust OCR settings (optional)
  4. Process - Run OCR on the document
  5. Download - Get your searchable PDF

Language Support​

Stirling-PDF supports OCR in 100+ languages including:

Common Languages​

  • English - eng
  • Spanish - spa
  • French - fra
  • German - deu
  • Italian - ita
  • Portuguese - por
  • Russian - rus
  • Chinese (Simplified) - chi_sim
  • Chinese (Traditional) - chi_tra
  • Japanese - jpn
  • Korean - kor
  • Arabic - ara
  • Hindi - hin
  • Dutch - nld
  • Polish - pol
  • Turkish - tur
  • Vietnamese - vie
  • Thai - tha

Multiple Languages​

If your document contains multiple languages, you can select multiple language packs for better accuracy.

Example: A document with English and Spanish text should have both eng and spa selected.

OCR Options​

Layout Preservation​

Options:

  • Preserve Original Layout - Maintains original page structure, formatting, and layout
  • Simple Text Layer - Adds searchable text without preserving complex formatting
  • Clean Text Only - Extracts text without any layout preservation

Recommendation: Use "Preserve Original Layout" for documents where visual structure matters (forms, tables, multi-column layouts).

OCR Quality Settings​

Options:

  • Fast - Quick processing, good for clean scans
  • Balanced - Good quality with reasonable speed (recommended)
  • Best - Maximum accuracy, slower processing

Preprocessing Options​

Improve OCR accuracy by preprocessing images:

  • Auto-rotate - Automatically detect and correct page orientation
  • Deskew - Fix slightly tilted/skewed scans
  • Despeckle - Remove noise and artifacts from scans
  • Remove Background - Clean up paper texture and shadows
  • Enhance Contrast - Improve readability of faded text

Tip: For poor quality scans, enable multiple preprocessing options.

Best Practices​

For Best OCR Results​

  1. Use high-quality scans

    • 300 DPI or higher recommended
    • Higher resolution = better accuracy
    • Minimum 150 DPI for acceptable results

  2. Clean, clear images

    • High contrast between text and background
    • Minimal shadows or stains
    • Sharp focus, not blurry

  3. Correct orientation

    • Text should be right-side up
    • Use auto-rotate if unsure

  4. Select correct language

    • Choose all languages present in document
    • Wrong language = poor accuracy

  5. Preprocess poor scans

    • Enable deskew for tilted pages
    • Use despeckle for noisy scans
    • Enhance contrast for faded text

Document Types​

Works Best With:

  • ✅ Scanned documents (text documents, contracts, letters)
  • ✅ Photos of documents taken with good lighting
  • ✅ Clean screenshots of text
  • ✅ Printed text (books, magazines, reports)

Challenging Cases:

  • ⚠️ Handwritten text (limited accuracy)
  • ⚠️ Stylized or decorative fonts
  • ⚠️ Very small text (< 8pt font)
  • ⚠️ Low resolution images
  • ⚠️ Heavily compressed or artifacted images

Common Issues​

"No text recognized"​

Possible Causes:

  • Wrong language selected
  • Image quality too poor
  • Text too small or blurry
  • Extreme skew/rotation

Solutions:

  • Verify correct language pack selected
  • Use higher quality scan
  • Enable preprocessing options
  • Check document orientation

"Poor accuracy / Garbled text"​

Possible Causes:

  • Low quality scan
  • Wrong language selected
  • Unusual font or formatting
  • Background interference

Solutions:

  • Increase scan resolution
  • Select multiple language packs
  • Enable despeckle and contrast enhancement
  • Clean up document before scanning

"Processing takes too long"​

Possible Causes:

  • Large document (many pages)
  • High resolution images
  • "Best" quality setting
  • Multiple preprocessing options

Solutions:

  • Process in smaller batches
  • Use "Balanced" quality setting
  • Reduce resolution if very high
  • Disable unnecessary preprocessing

Technical Details​

OCR Engine​

Stirling-PDF uses Tesseract OCR, an industry-standard open-source OCR engine originally developed by HP and now maintained by Google.

Key Features:

  • Over 100 languages supported
  • Multiple output formats
  • Layout analysis and preservation
  • Character and word confidence scores

Processing Steps​

  1. Image Analysis - Detect page layout and text regions
  2. Preprocessing - Apply selected image enhancements
  3. Text Recognition - Recognize characters using language models
  4. Layout Reconstruction - Preserve original formatting
  5. PDF Generation - Create searchable PDF with text layer

Output Format​

OCR produces a PDF with embedded text layer:

  • Original image preserved (visual appearance unchanged)
  • Invisible text layer added on top
  • Text is searchable and selectable
  • Layout matches original document

Configuration​

Installing Language Packs​

By default, Stirling-PDF includes common language packs. To add additional languages:

Docker:

# Install additional language packs
RUN apt-get update && apt-get install -y \
    tesseract-ocr-ara \
    tesseract-ocr-chi-sim \
    tesseract-ocr-jpn

See: OCR Configuration Guide for detailed setup instructions.

Environment Variables​

# OCR Settings
system:
  ocr:
    enabled: true
    languages: "eng,spa,fra,deu"  # Default languages
    pageSegmentationMode: auto

Use with Other Tools​

Common Workflows​

OCR → Convert to Word

  1. Run OCR to make document searchable
  2. Use Convert to export to DOCX
  3. Edit document in Word

OCR → Search & Redact

  1. Run OCR to add text layer
  2. Use search to find sensitive information
  3. Use Redact to remove it

OCR → Extract Data

  1. Run OCR on scanned forms/invoices
  2. Use PDF to CSV to extract tables
  3. Import data into spreadsheet

Scan → OCR → Compress

  1. Scan documents to PDF
  2. Run OCR to make searchable
  3. Use Compress to reduce file size

API Usage​

Perform OCR programmatically via API:

curl -X POST http://stirling-pdf:8080/api/v1/ocr/pdf \
  -F "fileInput=@scanned.pdf" \
  -F "languages=eng+spa" \
  -F "sidecar=false" \
  -F "deskew=true" \
  -F "clean=true" \
  -F "cleanFinal=true" \
  -F "ocrType=force" \
  -F "ocrRenderType=hocr" \
  -o searchable.pdf

Parameters:

  • languages - Language codes (+ separated)
  • sidecar - Generate separate text file
  • deskew - Fix tilted pages
  • clean - Remove noise
  • cleanFinal - Final cleanup
  • ocrType - skip, force, or auto
  • ocrRenderType - Output format

See API Documentation for complete endpoint reference.

  • Convert - Convert OCR'd PDFs to Word, text, or other formats
  • Compress - Reduce file size after OCR
  • Multi-Tool - Chain OCR with other operations
  • Auto-Rename - Rename files based on OCR'd content

Summary​

Stirling-PDF's OCR tool provides:

✅ 100+ language support - Recognize text in any language ✅ Layout preservation - Maintain original document formatting ✅ Batch processing - OCR multiple files at once ✅ Preprocessing options - Enhance poor quality scans ✅ Industry-standard engine - Tesseract OCR with proven accuracy ✅ API access - Automate OCR workflows

Perfect for digitizing scanned documents, making PDFs searchable, and extracting text from images!