Image Watermarking Robust to Print and Generation Copy

In general, after a watermarked image is printed, copied several times and scanned, the watermark usually cannot be extracted rightly, as well as distorted by noise. To solve this problem, in this paper, two watermarks were embedded into the image. The first one is a periodic watermark. The second one is the real watermark message. These two watermarks are embedded into the image by modifying the gray value of each pixel. The cover image is not needed when the watermark is detected. Both of these two watermarks are estimated exploiting the properties of HVS (human visual system) and extracted through the same key-dependent 2D sequences. Experiments results show that the scheme can resist print, generation copy and scan process