How harsh3029/digital-watermarking works
Project type and purpose
• This is a research‐oriented image processing pipeline (an open‐source tool / case study) for embedding and extracting invisible watermarks in digital images using visual cryptography.
• The four Python scripts each implement one stage of the pipeline:
• ownership_share_generator.py – builds the ownership share from the original watermark and host image
• master_share_generator.py – extracts master shares from possibly stolen/modified images
• watermark_generator.py – reconstructs the watermark by combining master + ownership shares
• template_match_res.py – computes accuracy of the reconstructed watermark
File structure
• Top‐level:
• Code/ – all processing scripts
• images/ – input sample images and generated outputs (master_images, stolen_images, regenerated_watermarks)
• LICENSE, README.md
• No services or databases: it’s a pure local pipeline.
• The four scripts execute in sequence (or independently) to form the full watermark cycle.
README
• Confirms project goal (visual cryptography watermarking).
• Lists run order and how each script interacts with image assets.
1. Input Data
– Original host image(s)
– Binary watermark image
2. Ownership Share Generator (Script 1)
3. Master Share Generator (Script 2)
4. Watermark Extractor (Script 3)
5. Accuracy Evaluator (Script 4)
6. Output Data
– Ownership share file
– Master share images
– Reconstructed watermarks
– Accuracy metrics
– Watermark + Host Image → Ownership Share Generator → Ownership Share
– For each stolen/modified Image + Ownership Share → Master Share Generator → Master Share
– Ownership Share + Master Share → Watermark Extractor → Reconstructed Watermark
– Reconstructed Watermark + Original Watermark → Template Matcher → Accuracy Score
c. Architectural patterns / principles:
– Pipeline pattern: clearly segmented processing steps
– Data‐driven flow: each script reads from images/ and writes back
– Single responsibility: each script has one discrete function
d. Key technologies & libraries:
– Python 3.x
– Image-processing libs (Pillow or OpenCV)
– File-system I/O for image assets
– Any pseudo-random generator module for point selection