【作者】 舒南飞；

【导师】 牛少彰；

【作者基本信息】 北京邮电大学 ， 应用数学， 2010， 硕士

【摘要】 随着网络的快速发展和多媒体数字作品的广泛使用,数字作品易于传播、分发和高质量复制等特点,使得获取数字作品的高质量拷贝越发的容易。致力于数字作品版权保护的数字水印技术作为一种数字权益保护的强有力手段的同时,也迎来了不断的挑战。能够抵抗打印扫描的数字水印算法,应用于打印扫描环境中,将能用于解决印刷作品的版权保护问题。本文主要研究了数字图像作为原始作品,在经历打印过程后,如何对其印刷作品进行版权保护的问题。设计能够抵抗打印和扫描的数字水印算法,在数字作品印刷前进行嵌入水印信息,通过扫描其打印后的印刷作品,从中提取水印信息,可以实现数字作品在印刷过程中的权益保护。本文在介绍了数字水印研究的相关基础知识后,给出了本文研究的理论意义和实际价值。在分析了数字图像在打印和扫描过程中经历的非线性失真和几何失真的主要攻击后,以寻找图像打印扫描前后不变的数字特征为指导思想实现水印信息的嵌入。通过分析图像分块DCT域系数在打印扫描前后的变化,得到了图像分块DCT系数的正负号关系的不变量。基于这个不变的数字特征,本文提出了种基于图像DCT域的鲁棒性水印和脆弱性水印共存的双水印算法。该双水印算法中的鲁棒性水印是通过修改同类DCT域系数的正负系数数量的差值关系实现信息的嵌入。实验证实,该鲁棒性水印具有较强的抵抗打印扫描攻击的性能。双水印算法中脆弱性水印的嵌入,丰富了该水印算法的职能,使其可以用于对鲁棒性要求较高和需要确定篡改定位的数字图像的网络传播中。基于数字图像在打印扫描后其语义信息不变的事实,通过在其D岍域的低频部分中提取该特征的数字表示,消弱了打印扫描过程中几何攻击对数字图像的影响。针对该不变的数字特征,提出了一种基于图像特征的抗打印扫描数字水印方案,通过对图像的DWT域低频系数进行分块聚类,再根据相邻块的类别差异,获得能够表示图像特征的极性表。对于待嵌入的水印信息,将其置乱后与图像的极性表异或得到密钥图像,实现水印信息的嵌入。嵌入过程中水印：的置乱参数和密钥图像等秘密参数决定了该水印方案的安全性。实验结果表明,在保证一定嵌入量,已知嵌入过程的秘密参数的情况下,嵌入的水印能够准确提取,且方案性能在安全性方面上优于相似方案,能够实现对印刷作品的版权保护。就鲁棒性数字水印算法整体性能的提升,本文给出了数字水印算法的鲁棒性、不可感知性和水印容量三者之间性能最优化的解决思路,在满足一定的不可感知性和水印容量要求的条件下,使数字水印具有更强的鲁棒性。最后,在总结本文的主要解决问题和取得成果后,整理了抗打印扫描数字水印研究中存在问题,并对其今后研究的方向和应用前景进行了展望。更多还原

【Abstract】 With the rapid development of Internet and the widespread use of multimedia digital works, the easy-to-spread distribution and high-quality reproduction and other special features of digital works, making access to high-quality copies of digital works becomes increasing easy. The technology of Digital watermarking aiming at the protection of digital work is a powerful mean of digital right protection, also gets consistently challenge. Print-scan resistant digital watermarking algorithm, used in the environment of print and scan, will be able to solve the issue of copyright protection in the printed works.In this paper, the main target is that the digital images as the original works in the experience of printing process, how to protect the copyright of printed works. Design the print-scan resistant digital watermarking algorithm, embedding watermark before printing digital works, by scanning the printed works to extract the watermark information, so as to realize the rights protection of digital works in the printing process.After describing the basic knowledge of digital watermarking, the paper shows its theoretical significance and practical value. After analyzing the digital image in the printing and scanning process through the main attack of non-linear distortion and geometric distortion, gives the idea by modifying the invariant mathematical feature to achieve the embedding of watermark as the guiding ideology.The paper finds that the sign of the DCT coefficiences of image block is invariant by analyzing the change about the block DCT coefficiences before and after print-scan. Based on this, the paper proposes a double watermarking algorithm based on DCT domain that the coexistence of robust and fragile digital watermarking can be achieved. The robust watermark in the double watermarking algorithm use the invariant feature after print-scan, the relationship between the number of positive and negative coefficients of the same DCT efficiencies class, accomplishes the embedding of watermark by modifying the difference relationship. Experiments demonstrate that the robust watermark can resist the print-scan attack. The fragile watermark in the algorithm enriches the function of the algorithm, making it do well in the Internet environment that needs strong robustness and can determines the location of the digital image tampered with the network transmission.Based on the fact that the unchangeable semantic information after print-scan, the digital presentation of the feature extracted from the low-frequency part of the image’s DWT domain, can reduce the effect of the geometrical attack during print-scan process. Accordding to this invariant feature, the paper proposes a digital watermarking scheme that based on image feature. It calculates the low-frequency wavelet coefficients of image, divides it by block and clusters, and according to the difference between the adjacent blocks, gets the polarity table of the image. After scrambling the watermark image to be embedded, combine the polarity table and scrambled watermark to get a key image by using XOR operator, accomplishing the embedded process. The security of the scheme is depended on the scrambling parameters and the key image, etc. Experiment result shows that, when certifying some amount of the bits embedded, the watermark can be extracted accurately, and better than other similar scheme, do implement the copyright protection of printed work.Then, the paper gives the solution idea of achieving the optimum performance among robustness, imperceptibility and capacity of digital watermarking algorithms. When meeting certain required conditions about perceptibility and watermark capacity, digital watermark can reach stronger robustness.Finally, after concluding the main solved problems and achievement, summaries the remained problem in the research of print-scan resistant digital watermarking, and prospects the direction for future research and application.更多还原