Efficient Robust Watermarking Based on Structure-Preserving Quaternion Singular Value Decomposition.

Area of Science:

• Digital image processing
• Information security
• Multimedia security

Background:

• Quaternion Singular Value Decomposition (QSVD) is a robust digital watermarking technique.
• Existing QSVD methods suffer from high complexity, limiting real-time application, invisibility, and robustness.

Purpose of the Study:

• To develop a novel, efficient, and robust QSVD-based digital watermarking scheme.
• To address the complexity and performance limitations of current QSVD watermarking techniques.

Main Methods:

• Introduced a new real structure-preserving QSVD algorithm.
• Implemented blind secret information transmission using coefficient pair selection and adaptive embedding.
• Utilized normalized cross-correlation for selecting highly correlated coefficient pairs.
• Employed a Logistic chaotic map for secure key generation.

Main Results:

• The proposed scheme demonstrates high efficiency and fidelity with low distortion.
• Numerical experiments show resistance to various image attacks, including large-size 8-color binary watermarks and QR codes.
• The method achieves superior real-time performance, invisibility, and robustness compared to state-of-the-art approaches.

Conclusions:

• The novel QSVD watermarking scheme effectively overcomes complexity issues.
• The scheme offers significant improvements in real-time processing, invisibility, and robustness.
• The use of coefficient pair selection, adaptive embedding, and chaotic maps enhances security and performance.