Research On Blind Detection Digital Watermarking Technology Against Geometric Attacks

Digital watermarking technique which is a new method for information security and copyright protection is researched widely. Although many methods are proposed, most of them still can't withstand geometrical attacks. So the researches of digital watermarking technology that are resistant against geometrical distortions are becoming more important. This paper takes still images as examples and three new digital watermarking methods are proposed.First of all, a digital watermarking method based on the neighboring relation in the contourlet transform domain is proposed. Firstly, the original image is transformed by contourlet transform. Then the watermarking is embedded in the selected detail subband according to average value of the neighboring symbol. The method is simple and the original image is not needed in the detection phase. And the digital watermarking is a binary image. The experimental results show that the scheme can well withstand attacks such as JPEG lossy compression, additive noise, gaussian filtering and sharpening.Secondly, a robust watermarking technology applying the krawtchouk moment and wavelet transform is proposed. Firstly, the subband coefficients are extracted after discrete wavelet transformation. Then, the lower order krawtchouk moments of the low frequency subbands coefficients are computed. The mean of several functions of the moments is used to judge whether the watermarking is existent. The process of encoding and decoding is simple, just computing a few lower order moments. Experimental results show that the scheme is simple, effective, and can well withstand such geometrical attacks as rotation, scaling, shearing, combined attack, and JPEG lossy compression.Finally, a robust watermarking technology based on the trace transform is proposed. The features of the image are firstly extracted by the trace transform. Then the message is registered at the fair third party in order to gain the timestamp. In the end, we can use the timestamp to prove who the real owner is. The experimental results show that the scheme can also well withstand such geometrical attacks as rotation, scaling, shearing, combined attacks, affine transform, removing lines, filtering and JPEG lossy compression.