A grayscale watermark technique using sub-sampling and singular value decomposition

In this work, we propose a robust watermark technique to embed a grayscale watermark into a host image. In the embedding process, the host image is decomposed into four sub-images using sub-sampling and each sub-image are divided into blocks according to the size of the watermark. Two pairs of blocks are taken considering one block (similar position) from each sub-image. The watermark pixel is embedded into each pair of block by modifying the largest singular value of each block. Before embedding, the watermark is scrambled by Arnold transform and then, normalized by a watermark depended secret key to enhance the security level. The experimental result shows that the proposed scheme is robust against different attacks such as JPEG compression, salt and pepper noise, cropping, rotation etc. The proposed scheme is also compared with another grayscale watermark scheme. Our proposed scheme shows better robustness under attacks.