Robust multi-modal watermarking using visually encrypted watermark

The paper presents a robust and secure multi-modal strategy for watermarking of the digital images. In this approach, the image is divided into blocks. The blocks are first permuted according to a pseudorandom key. All the blocks of the original image are statistically analyzed and the watermark is embedded into only those blocks in which the standard deviation of the image pixel intensities is greater than a predefined threshold. Leaving some of the blocks un-watermarked decreases chances of blind detection by the attacker due to probability of false detection from the non-watermarked blocks. The watermark is embedded in the blocks using randomly selected transformation strategies using any of the three transforms, viz. Discrete Cosine Transform, Discrete Fourier Transform, and Discrete Wavelet Transform. Each block is independently watermarked with one unique share of original watermark generated through threshold visual cryptography method. Encrypting the watermark before embedding makes the watermark detection and removal very difficult making it secure. Experimental results depict the robustness of the algorithm against prevalent image processing attacks like rotation, scaling, compression, noise addition and filtering, etc.