A Novel Watermarking Algorithm Resist to Geometrical Attacks

In this paper we present a novel watermarking algorithm based on feature points and integer discrete wavelet transform (IDWT) against geometric rotation and scaling. Geometric distortions are critical factors which hamper the application of the image digital watermarking in that most of the watermarking technique would fail after the image has been experienced geometric transform. The proposed scheme enhances the security of the embedded watermarks by utilizing Arnold transform. It embeds the watermark in such a way that it partially changes the arithmetical compliment of the deep low frequency wavelet coefficient according to the watermark's characteristic. Particularly, the method tries to correct the geometric distortions by comparing the Hausdorff distance of the wavelet feature sets before and after experiencing the distortion. As will be shown in the paper, the method is stable and robust while against highly intensive geometric distortion.