Palm Leaf Manuscript Enhancement and Line Segmentation

Classical image processing pipeline for palm leaf manuscript enhancement and line segmentation using Python and OpenCV.

Objectives

• Enhance manuscript readability
• Reduce background interference
• Improve binarization quality
• Perform text line segmentation

Methodology

1. Image Preprocessing

• Grayscale conversion
• CLAHE contrast enhancement
• Bilateral filtering
• Background normalization

2. Binarization

• Gaussian smoothing
• Otsu thresholding
• Morphological operations

3. Line Segmentation

• Horizontal projection analysis
• Region detection
• Segmentation refinement

Tools & Libraries

• Python
• OpenCV
• NumPy
• Matplotlib

Key Insight

The project demonstrated that improving preprocessing and binarization quality had a greater impact on segmentation accuracy than modifying segmentation algorithms.

Output

The pipeline enhances degraded manuscript images and enables projection-based line segmentation for computational manuscript analysis.

Results

Input Manuscript

Segmentation Output

Limitations

• Segmentation accuracy depends on manuscript image quality.
• Severe degradation and illumination variation may affect line detection.
• Highly noisy regions may produce merged segmentation outputs.

Future Scope

• OCR integration
• Deep learning-based manuscript restoration
• AI-assisted document enhancement
• Computational manuscript transcription systems

Author

Mabbu Ketan Prakash Reddy
M.Sc Bioinformatics | SVIMS