基于改进脊波变换的抗攻击数字水印算法

Digital image watermarking algorithm resisting attacks based on improved ridgelet transform

针对小波变换域不是图像轮廓边缘特征最佳表示的问题,在研究小波变换的基础上,利用脊波变换比小波变换更适合表示图像轮廓边缘特征的特性,结合人类视觉系统特性和奇偶量化算法改进脊波变换算法,提出了一种基于改进脊波变换的数字水印算法。将原始图像均匀分割,对各子块执行脊波变换,再结合人类视觉系统特性选择恰当的脊波系数,利用奇偶量化算法嵌入一维水印序列,最后作脊波反变换得到含水印图像。仿真实验结果表明,算法不仅具有很好的透明性,而且在抵抗攻击时表现出较好的鲁棒性。通过与小波变换算法的性能比较,本文算法在抗加噪、滤波和剪切的攻击中鲁棒性较强。

Wavelet transform is not the best method to describe contour feature. Ridgelet transform is more suitable to describe the characteristic of image contour edge than wavelet transform. This paper used the characteristics of human vision system and odd-even quantization algorithm to improve ridgelet transform. It proposed a watermarking embedding algorithm based on im- proved ridgelet transform. Firstly, it segmented a cartier image evenly. Secondly, it made each sub-block ridgelet transform. Thirdly, it confirmed the appropriate ridgelet coefficients according to the characteristic of human vision system. Each sub- block was embedded the one-dimensional watermarking sequence using odd-even quantization algorithm. Finally, it achieved the watermarked image by inverse ridgelet transform. The simulation results demonstrate that the embedded watermark can have good invisibility and robustness for the common image processing, especially noise, filtering and cropping. It compared with the wavelet transform algorithm, this algorithm has stronger robustness against noise, filtering and cropping attack.