Robust Digital Image Watermarking Using DWT, Hessenberg, and SVD for Copyright Protection

Abstract

With the rapid advancement of digital technology, protecting visual content against misuse, unauthorized copying, and forgery has become increasingly important. This study proposes a robust and secure image watermarking method that synergistically integrates Discrete Wavelet Transform (DWT), Hessenberg Decomposition (HD), and Singular Value Decomposition (SVD) to enhance both imperceptibility and robustness against common image attacks. The method involves embedding the watermark into the low-frequency coefficients obtained through DWT, leveraging HD for structural stability, and using SVD to embed the watermark into the singular values, ensuring minimal perceptual distortion. The approach is designed to withstand a wide range of attacks, including filtering, noise addition, cropping, and lossy compression, while maintaining high fidelity of the watermarked images. The implementation was tested on multiple grayscale images, such as Lena, Baboon, and Cameraman. The robustness and transparency of the watermarking technique were evaluated quantitatively using metrics including Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), and Normalized Correlation (NC), under various attack scenarios. Results indicate that the proposed method achieves high PSNR and SSIM values, demonstrating excellent imperceptibility, while maintaining a strong resistance to image distortions, as evidenced by the NC values remaining close to 1 after attacks. The experimental results confirm the method’s capability to preserve the embedded watermark reliably, with minimal perceptible differences in the visual quality of watermarked images.