A robust digital image watermarking scheme for content protection

In this paper, a blind image watermarking scheme is proposed for resist the attacks such as collages, cutouts, etc. Traditional image watermarking technology cannot resist such attacks, because the synchronization information is broken. We use the noise template as the watermark, and the watermark can be detected with a high accuracy from part of the watermarked image. We also propose an improved weight template, which can enhance the imperceptibility of the watermark. In the watermark detection process, the autocorrelation of the high frequency component is necessary. Because of the existence of the attack, the first autocorrelation contains much disturbing peak points. In order to enhance the feature and suppress the interference, we propose the second autocorrelation on the result of the first autocorrelation. Next, a simple and effective method is proposed for detection and decision. In addition, some tricks are proposed to get a higher detection accuracy. Experimental results demonstrate that our scheme outperforms traditional schemes in resisting collage, cutout and image synthesis, and our scheme has a high accuracy, and the proposed watermarking scheme has a strong robustness to common image processing operations.