Histogram based watermarking algorithm resist to geometric attacks

The image's histogram use the statistical features of an image and is regardless of the specific distribution of image pixels. With this feature, the histogram-based image watermark algorithm can avoid the desynchronization that geometric attacks bring to image and is robust to geometric attacks. Typical histogram algorithms can resist translation and crop attacks, but the ability to resist rotation, scaling and JPEG compression attacks need improvement. In this paper we divide a group into more bins, give triple relationships of bins. We use election mechanism to improve detection efficiency. Experiment results show our algorithm have higher detection correct rate in rotation, scaling, JPEG compression attacks than previous algorithm.