【作者】 王庆梅；

【导师】 刘凤玉；

【作者基本信息】 南京理工大学 ， 计算机应用， 2004， 博士

【摘要】 随着计算机和网络技术的飞速发展，信息安全问题愈来愈受到人们的关注，密码技术在文本信息的加密方面非常有用，但随着多媒体技术的发展，多媒体信息的版权保护、真伪鉴别、盗版跟踪等问题逐渐浮现水面。密码技术在解决这些问题就无能为力了，而目前国内外正研究的数字水印技术是解决这一问题的良好途径。 数字水印的研究对象主要包括静止图像、视频和声频信息。重点对静止图像进行研究。静止图像的文件格式有很多，常见的有BMP，JPEG，PCX，GIF，TIFF等。由于它们之间可以相互转换，本文仅考虑BMP和TIFF格式的文件。 本文在前人工作的基础上，主要创新性工作为： 1．利用双随机矩阵的特性对水印信息进行预处理，可以把水印每一点信息分散嵌入到载体图像的多个点上。双随机矩阵与模运算的有效结合使得公钥数字水印算法成为可能。双随机矩阵的引入可以保证水印信息具有基于DCT水印算法的一切优点的基础上，还可以正确检测出切割攻击。 2．提出了基于多签字体制的公钥数字水印嵌入与提取方案，可以实现水印的嵌入密钥掌握在两个不同的用户手中，任何掌握公钥的人都可以正确提取出待测图像中含有的水印信息。 3．给出了具有多个仲裁者的数字水印嵌入与验证方案，其主要思想是将水印提取密钥分拆为数个密钥片段，并由不同的用户或仲裁者分别保管。本方案可以实现秘密共享，以保证只有若干个人同时在场的情况下才能启用。 4．有效的将背包密码体制和模运算结合到数字水印技术中，使得攻击者试图通过穷举攻击破译水印信息时将面临着双重不可求问题，即：单向陷门背包问题和解离散对数问题。 5．利用图像分块的灰度差异值判定水印信息的嵌入位置，由于水印信息嵌入到了纹理复杂度因子比较大的块中，这样就增强了水印信息的不可见性。实验表明，本算法较好的保持了原始图像的品质，对非几何攻击和有损压缩具有较好的稳健性。 6．将图像融合的思想引入到数字水印技术中来，得到一种适于载体图像为彩色图像的数字水印算法。算法可使载体图像的HSV变换法的高频信息较好的保留，也能够保证水印图像的良好的空域和频域的局域性。更多还原

【Abstract】 With the widespread distribution of digital media contents, the protection of the intellectual property rights has become increasingly important. Digital media offers several distinct advantages over analog media, such as high quality, easy editing, high fidelity copying. Digital information can be duplicated and distributed easily has led to the need for effective copyright protection tools. Cryptography can do nothing in resolving above problems, but digital watermarking which is one of the research focus in the academic world presents a candidate key to those questions.What digital watermarking technique mainly research on is still images, video and audio signals. The study object this file is still images which are most researchers discussed on because of its convenience. The focus of this dissertation is the theories and algorithms of digital image watermarking based on many research works of a lot of academicians and experts. The key points of research addressed in this proposal are as following:1. Double stochastic matrix can map every watermark image points to the corresponding point of the host image. In this way, watermarks can be distributed all over the original image and it’s difficult for attackers to remove or change watermarks even if he has the full knowledge of the verification algorithm owing to the deficiency of the secrete key. An asymmetrical watermarking algorithm is also proposed in this file that combines this kind of matrix with modulations.2. A public digital watermarking scheme based on multi-signature is presented. The main idea of this scheme is that secrete keys used to embedding are kept secrete by different person, while anybody holding the public key can pick-up watermarking information from watermarked images.3. Give out a digital watermarking embedding and detection scheme with several arbitrators. The key used to detecting watermarks is divided into several key segments, and kept by different users or arbitrators. This scheme realizes secrete sharing and guarantees the detection process can be done only when there are several different persons.4. Put forward a digital watermarking combined Merkle-Hellman knapsack system with modular arithematic. If an attacker want to destroy thewatermarks he will face to dual NP: the knapsack problem and the discrete logarithm problem.5. Bring forward a new digital watermarking scheme in which the embedding position is determined by the different value of the gray of the images blockwise. And compute the DCT transformation to those selected blocks, and then embed watermarks into the middle coefficients. It can give little degrade to the original image, and can effectively resist non- geometric attack and lossy compression.6. Introduce the image confusion techniques into digital watermarking system, and obtains an embedding algorithm which is adapt to colored images.更多还原