Robust image watermarking method based on DWT-SVD-Schur decomposition
Xuejian Yang, Baokun Hu · 2023
Network communication plays a crucial role in modern life. The ease of sharing and storing images on the Internet enhances user convenience. However, the transmission process exposes images to the risk of theft or unauthorized duplication by malicious actors, leading to copyright disputes. Addressing these disputes has made digital watermarking a focal point in the realm of copyright protection. This paper introduces an innovative approach that combines Discrete Wavelet Transform (DWT), Singular Value Decomposition (SVD), and Schur decomposition. The proposed method initiates by applying R-level DWT to the host image, employing Schur decomposition on the low-frequency sub-band, and subsequently subjecting the obtained diagonal matrix to SVD processing. Simultaneously, SVD processing is applied to the watermark image. Ultimately, the optimal scaling factor is chosen to embed the watermark image into the host image. Watermark embedding and extraction are inherently opposing processes. This study utilizes Peak Signal-to-Noise Ratio (PSNR) to assess the invisibility of the watermark and the Normalized Correlation (NC) value to evaluate the robustness of the watermarked host image. Experimental results demonstrate that the method proposed in this paper exhibits high invisibility and formidable robustness against various watermark attacks. Furthermore, a comparative analysis with methods from other papers reveals the superior resistance of the proposed method against watermark attacks.