Li Jianzhong, Zhang Xiangsu, Liu Shou, Ren Xuechang
Department of Physics, Xiamen University, Xiamen, Fujian 361005, China.
Appl Opt. 2009 Sep 10;48(26):4858-65. doi: 10.1364/AO.48.004858.
We present a new adaptive blind watermarking scheme using a gray-level computer generated hologram in the discrete wavelet transform domain. By utilizing an improved fuzzy clustering technique and human visual system , the watermark can be adaptively embedded according to block classification. To keep imperceptibility and robustness, a novel iterative embedding algorithm is adopted to change the to-be-embedded coefficients. Compared with the standard Fuzzy c-means (FCM) clustering, the suggested improved FCM (IFCM) converges more quickly and can avoid local optimum effectively. The experimental results demonstrate that the proposed scheme provides good robustness to withstand different kinds of common attack. Compared with other methods, the proposed method has the distinct advantage of better robustness to a JPEG compression attack.
我们提出了一种新的自适应盲水印方案，该方案在离散小波变换域中使用灰度级计算机生成全息图。通过利用改进的模糊聚类技术和人类视觉系统，水印可以根据块分类进行自适应嵌入。为了保持不可感知性和鲁棒性，采用了一种新颖的迭代嵌入算法来改变待嵌入系数。与标准模糊C均值（FCM）聚类相比，所提出的改进FCM（IFCM）收敛更快，并且可以有效避免局部最优。实验结果表明，所提出的方案具有良好的鲁棒性，能够抵御不同类型的常见攻击。与其他方法相比，该方法在抵抗JPEG压缩攻击方面具有明显更好的鲁棒性优势。
