Digital Image Watermarking using Dual-scrambling and Singular Value Decomposition

This paper proposes an algorithm to perform robust digital watermarking through image scrambling and singular value decomposition (SVD). Both the watermark and the carrier are images, and the proposed algorithm aim to ensure the robustness and perceptual invisibility of the embedded watermark. The proposed algorithm first performs a scrambling encryption on the watermark and carrier images. The scrambled image is then partitioned and each partition is decomposed using SVD. A quantization parameter is defined to modify the singular value of each partition and the watermarks are inserted into the feature domains. Experimental results show that the proposed algorithm is highly effective in protecting the invisibility of the inserted watermarks. The algorithm is also remarkably robust to image processing methods such as JPEG compression and filtering, and to malicious attacks such as clipping, rotation and scaling.