中文摘要：

数字水印技术已成为多媒体信息安全领域的一个研究热点.本文对宿主图像和水印图像DCT系数的幅值分布情况和相互关系进行了理论上的分析,得出二者的分布可以通过一个自适应关系矩阵进行拟合的理论结果,并给出了计算自适应关系矩阵的快速算法.在此基础上,以自适应关系矩阵为边信息,提出了一种基于宿主图像和水印图像DCT系数分布的自适应边信息水印算法.该算法将灰度图像作为水印信息,利用自适应关系矩阵调节水印图像DCT系数的样本值,并用调节后的样本值替换伪随机选择出的宿主图像中嵌入位置的DCT系数,实现水印嵌入.自适应边信息的引入,使水印算法获得了尽可能多的优化.与同类方法比较,所提出的算法透明性好,鲁棒性强,安全性高,对常见的几何、滤波、噪音、压缩等多类攻击性能很好,水印提取不需要原始宿主图像.实验结果验证了方法的有效性.另外,本文所提出的自适应矩阵及其快速计算方法,还可应用于其它类似的分布拟合过程.

英文摘要：

The digital watermarking technology has become a research hotspot in the field of multimedia information security. This paper examines the amplitude of the host image＇s and the watermarking image＇s DCT coefficlents,then gives a generally theoretical analysis on the distribution law and characteristics of the two groups＇ coefficients＇ amplitudes and the relation between them. The analysis draws a conclusion that the two distributions can be fitted by an adaptive relation matrix and puts forward a fast algorithm to compute the adaptive relation matrix. Then,taking the adaptive relation matrix as side information,an adaptive image watermarking scheme with side information is proposed. This scheme adopts a gray-level image as the watermarking,embeds the watermarking by means of modulating each DCT coefficient＇s value of the watermarking image according to the relation matrix and then using it to respectively replace the DCT coefficient＇s value at the relevant position,which is chosen pseudo randomly,of the host image. The side information can optimize the watermarking algorithm as much as possible. The experimental results show that ,compared with algorithms of the same kind ,the proposed algorithm can reach good transparency,high safety,and excellent robustness to common attacks such as image cropping,filtering ,noises and JPEG lossy compression. The watermarking can be extracted without original host image. Besides ,the adaptive relation matrix and its fast computation can be applied to similar numeric fitting as well.