This paper presents a robust color image watermarking scheme based on Tensor Singular Value Decomposition (Tensor-SVD) . This is designed to effectively exploit the strong inter-channel correlation of RGB color images. By modelling the RGB components as a third-order tensor, the proposed approach enables joint spatial-spectral processing, which is not achievable using conventional 2D SVD-based methods. To enhance imperceptibility and preserve image integrity, a one-level Integer Wavelet Transform (IWT) with integer-to-integer mapping is applied independently to each RGB channel. Subsequently, corresponding wavelet sub-bands from the RGB channels (L R -L G -L B ) are combined to form a third-order tensor, and watermark embedding is performed by modifying dominant tensor singular components. This tensor-domain representation significantly improves robustness against a wide range of signal processing and geometric attacks commonly encountered in RGB images. The selection of effective embedding sub-bands is carried out experimentally based on their inherent resilience, ensuring an optimal trade-off between robustness and imperceptibility. The performance of the proposed method is evaluated using multiple metrics, including Normalized Correlation Coefficient (NCC) , Weighted Peak Signal-to-Noise Ratio (WPSNR) , Feature Similarity Index Measure (FSIM) , Peak Signal-to-Noise Ratio (PSNR) , and Structural Similarity Index Measure (SSIM) . Experimental results on standard color image datasets demonstrate an average NCC of approximately 0.9998 , SSIM of 0.9991 , PSNR of 47.63 dB , WPSNR of 46.77 dB , and FSIM of 0.9778 . Both subjective and objective evaluations confirm the superiority of the proposed method in terms of robustness and imperceptibility performance compared to different state-of-the-art watermarking techniques, making it well suited for copyright protection applications .
Barlaskar et al. (Fri,) studied this question.