Robust Reversible Watermarking With Invisible Distortion Against VAE Watermark Removal

Area of Science:

• Digital Image Processing
• Information Security
• Computer Vision

Background:

• Orthogonal Moment-based Robust Reversible Watermarking (OM-RRW) is vital for intellectual property protection.
• Existing OM-RRW methods embed watermarks in visually sensitive areas, causing ring-like distortions and making them vulnerable to inpainting attacks.

Purpose of the Study:

• To develop an innovative strategy to eliminate visible distortions in OM-RRW.
• To overcome the inherent limitations of traditional OM-RRW methods.

Main Methods:

• Customizing embedding step sizes based on moment value stability differences to minimize distortion.
• Designing a texture-aware adaptive basis function fine-tuning strategy to adjust representation capabilities based on human visual sensitivity.

Main Results:

• The proposed method successfully embeds 128-bit watermarks with no visible distortions.
• The method minimizes the loss of robustness while eliminating ring-like artifacts.
• OM-RRW demonstrated satisfactory robustness against Variational Autoencoder (VAE) watermark removal attacks.

Conclusions:

• The novel strategy effectively eliminates visible distortions in OM-RRW, enhancing its practical applicability.
• The improved method offers a more secure and visually imperceptible solution for digital watermarking.