SITI BALQIES LAKASEK, Haitham Elhadi
The rapid growth of digital communication and social media platforms has increased the risk of unauthorized copying, tampering, and copyright infringement of digital images, making digital image watermarking an essential mechanism for ownership protection while preserving visual quality. Although existing wavelet-based watermarking techniques provide good invisibility and robustness against common signal processing attacks, their limited directional representation reduces resistance to geometric distortions such as rotation, scaling, and cropping. To address this issue, this paper proposes a novel blind image watermarking method based on the Non-Subsampled Shearlet Transform and Singular Value Decomposition. In the embedding stage, the host image is decomposed using a multi-scale and multi-directional transform, and watermark information is embedded into the high-frequency directional sub-band using stable singular values to enhance robustness. The watermark is extracted without requiring the original host image, satisfying the blind watermarking requirement. The proposed method is evaluated using the peak signal-to-noise ratio and structural similarity index for imperceptibility, as well as normalised correlation and bit error rate for robustness under various signal processing and geometric attacks. Experimental results demonstrate that the proposed approach achieves an improved balance between imperceptibility and robustness compared to conventional wavelet-based methods with PSNR value of 38 db, and NC value of 0.99 under JPEG2000 compression. The results show the proposed scheme outperform the existing schemes in terms of robustness under Gaussian low-pass filtering, salt-and-pepper noise, JPEG compression, and JPEG2000 compression.