一种新的基于伪Zernike矩的图像盲水印算法

抵抗几何攻击的鲁棒性是目前数字水印技术中的热点也是难点。提出一种新的基于伪Zernike矩的图像盲水印算法，首先计算图像归一化后的伪Zernike矩，然后选取部分合适的矩通过量化调制嵌入水印信息。水印提取时，利用伪Zernike矩的相位信息估计旋转角度进行几何校正，以提高矩的几何不变性。实验结果表明，本算法对于抗几何攻击尤其是旋转攻击具有很好的鲁棒性，同时也能抵抗常规的信号处理攻击。

Novel blind image watermarking based on pseudo Zernike moments

How to improve the robustness watermarking against geometrical attacks is still a difficult and open problem of watermarking researchers. This paper proposes a novel blind watermarking algorithm based on pseudo Zernike moments. The standard image is obtained by preprocessing the original image, then calculating the pseudo Zernike moments and their magnitudes are quantify-modulated to embed watermark. In watermark extraction, the watermarked image is geometric corrected and the rotation angle is estimated by using the phase information of pseudo Zernike moments. Experimental results show that the proposed watermarking algorithm is robust to geometric attacks especially rotating attack, as well as common signal processing attacks.