An Improved DCT-SVD-Based Image Watermarking Method for Optimized Imperceptibility and Resilience

This research proposes a resilient image watermarking technique that synergistically combines the Discrete Cosine Transform (DCT) and Singular Value Decomposition (SVD) to enhance visual transparency and resistance to common image manipulation attacks. The approach begins by applying DCT to convert the spatial image domain into frequency components, from which suitable frequency coefficients are identified for watermark embedding. SVD is then applied to modify the singular values of the DCT-domain coefficients, enabling watermark insertion in a way that is both imperceptible and robust to degradations. The technique is engineered to achieve an optimal compromise between perceptual invisibility and embedding durability, ensuring the watermark remains unnoticeable to human observers while withstanding challenges such as compression artifacts, additive noise, and spatial filtering. Experimental evaluations demonstrate improved efficacy in terms of visual quality—measured by Peak Signal-to-Noise Ratio (PSNR)—and robustness against a range of signal processing operations. This method represents a practical and effective solution for secure digital watermarking in multimedia content protection.