基于伪Zernike的归一化分形水印算法研究*

以归一化技术，分形编码技术及伪Zernike矩相关知识为基础，提出一种可有效抵抗几何攻击的鲁棒数字水印新算法。算法首先利用归一化技术和不变质心理论在图像中提取出重要区域；然后利用分形编码及设置的阈值将重要区域分成自相似性块和非自相似性块并计算自相似性块的伪Zernike矩，从中选出最鲁棒的矩；最后通过量化调制伪Zernike矩幅值将水印嵌入其中。仿真实验表明，算法不仅具有较好的透明性，而且对常规信号处理(滤波、锐化、加噪和JPEG压缩等)和几何攻击(全局仿射变换、局部失真等)均具有较好的鲁棒性。

New normalized fractal watermarking scheme based on pseudo-Zernike moments

This paper proposed a new image watermarking scheme robust to geometric attacks.Firstly, extracted significant region by using normalization and invariant centroid theory.Then, divided significant region into two groups:self-similarity and non-self-similarity blocks by fractal coding and pre-set threshold, picked up the robustest pseudo-Zernike moments among the pseudo-Zernike moments of the self-similarity blocks.Finally, embedded the watermark by quantizing the magnitudes of the robustest pseudo-Zernike moments. Experimental results show that the scheme is not only invisible and robust against common signals processing such as median filtering,sharpening ,noise adding, JPEG compression,etc, but also robust against the geometric attacks such as affine transform,local geometric distortion, etc.