pdf-ocr-confidence 0.1.0

Estimate OCR reading confidence for PDF files before running expensive OCR

PDF OCR Confidence

A Python library to estimate OCR reading confidence for PDF files before running expensive OCR models.

What It Does

Analyzes PDF documents and returns a confidence score (0-1) indicating how well OCR will perform, helping you:

• Route low-quality PDFs to expensive/robust OCR models
• Route high-quality PDFs to cheap/fast OCR models
• Skip OCR entirely for native text PDFs

Features

• ✅ Detects native text layer (instant 100% confidence, skip OCR)
• ✅ Image quality analysis (resolution, blur, contrast, noise)
• ✅ Text region detection and edge analysis
• ✅ Per-page and document-level confidence scores
• ✅ Configurable thresholds for routing decisions

Installation

pip install -e .

Quick Start

from pdf_ocr_confidence import analyze_pdf

# Analyze a PDF
result = analyze_pdf("document.pdf")

print(f"Confidence: {result.confidence:.2f}")
print(f"Recommendation: {result.recommendation}")  # "native", "cheap_ocr", "expensive_ocr"

# Access per-page details
for page in result.pages:
    print(f"Page {page.number}: {page.confidence:.2f} - {page.quality_metrics}")

How It Works

1. Native Text Detection: Checks if PDF has extractable text layer
2. Image Quality Analysis:
   • DPI/Resolution check
   • Blur detection (Laplacian variance)
   • Contrast analysis (histogram)
   • Edge density (text clarity)
3. Confidence Scoring: Weighted combination of metrics
4. Routing Recommendation: Based on configurable thresholds

Configuration

from pdf_ocr_confidence import analyze_pdf, ConfidenceConfig

config = ConfidenceConfig(
    expensive_ocr_threshold=0.4,  # Below this = expensive OCR
    cheap_ocr_threshold=0.7,      # Above this = cheap OCR
    sample_pages=3,               # Pages to analyze (None = all)
    min_dpi=150,                  # Minimum acceptable DPI
)

result = analyze_pdf("document.pdf", config=config)

Dependencies

• PyMuPDF (fitz) - PDF text extraction and rendering
• Pillow - Image processing
• numpy - Numerical operations
• opencv-python - Advanced image analysis

License

MIT