Statistical Features Based Image Watermarking Resisting to Geometrical Attacks

The aim of our robust technique using watermarking is to reconstruct watermark from geometric attacks. Various geometric transformations, such as rotation, scaling, cropping, shear, translation, resize, scattering, bending, and so on, provides the displacement of some of its picture elements. The existing watermarking algorithms countermeasures the attacks related to pixel positions. In our paper, we use the histogram of image based watermarking system. While the histogram, is invariant to scaling mathematically, resistant to cropping statistically and independent of pixel position. By modifying the histogram shape a PN sequence is generated and uses as key, which is computing from the Gaussian filtered low-frequency component of images. Finally, the proposed watermarking scheme is more resistant to geometric deformations, including cropping, filtering, scaling, JPEG compression, shear and random bending attacks.