efdc0dc886
- Added 5 kernel types: text, structure, summary, philosophical, semantic - Improved diagram type detection with content analysis - Added color analysis and grayscale detection - Enhanced philosophical keyword extraction - Added semantic relationship detection - Improved error handling for missing dependencies - Added comprehensive testing with text-rich test images - Enhanced metadata and tagging system Key improvements: ✓ Semantic relationship detection (source → target patterns) ✓ Enhanced philosophical content extraction ✓ Color analysis and grayscale detection ✓ Better diagram type classification ✓ Comprehensive metadata and tagging ✓ Improved error handling and dependency warnings Still requires OCR dependencies for text extraction: - pytesseract for OCR - pdf2image for PDF processing - Tesseract OCR engine (see issue #563)