Cover-Lossless Robust Image Watermarking Against Geometric Deformations

Area of Science:

• Digital Image Processing
• Information Security
• Computer Vision

Background:

• Cover-lossless robust watermarking aims to fully restore original images after watermark extraction, even under attack.
• Existing methods struggle with geometric deformations (rotation, scaling) due to pixel-position-dependent features.
• Robustness against common manipulations like JPEG compression and noise is often prioritized over geometric invariance.

Purpose of the Study:

• To develop a novel cover-lossless robust watermarking method resistant to geometric deformations.
• To enable complete restoration of the cover image without any loss when no attacks occur.
• To enhance watermarking robustness against both geometric transformations and content-preserving manipulations.

Main Methods:

• Embedding watermarks into low-order Zernike moments, which are invariant to scaling and rotation.
• Reversibly hiding distortion as compensation information for cover image restoration.
• Utilizing quantized error, watermarked error, and rounded error to minimize compensation data.

Main Results:

• The proposed method demonstrates mathematical invariance to image scaling and rotation.
• It exhibits robustness against interpolation errors and common image processing operations.
• Experimental results confirm reduced compensation information and strong watermarking robustness.

Conclusions:

• A new cover-lossless robust watermarking system against geometric deformations is achieved.
• The method effectively reduces compensation information while maintaining high robustness.
• The system ensures complete cover image recovery without loss in the absence of attacks.