Geometrically robust image watermarking based on image normalization

In image watermarking, geometric transformations are among the most challenging problems. Moment based image normalization has been used in watermarking. Embedding the watermark into the normalized image is the idea of this kind of algorithms. However, because the sizes of the normalized original image and the normalized geometrically attacked image are different, the synchronization of the embedding and the extracting is very difficult to be guaranteed. In this paper, we propose a watermarking algorithm based on image normalization. We choose a parameter in the normalized image as the watermark. And the watermarked image is the combination of the original image and a weighted nonlinear transformation of the original. Watermark detection is as simple as computing the parameter of the normalized version of the received image. The proposed algorithm has proved to be robust to many kinds of geometric manipulations that are performed as part of StirMark attacks.