基于静止图像的数字水印技术研究

【作者】 程兴国；

【导师】 王蔚然；

【作者基本信息】 电子科技大学 ， 通信与信息系统， 2005， 硕士

【摘要】 当前快速发展的新的IT 技术、电子商务及大量商用多媒体业务的涌现,使得各种多媒体数据的版权保护技术显得尤为重要。近年来,国外许多学者提出了一系列新的信息安全保护思想,特别是在知识产权保护、防篡改及信息内嵌式注释等领域提出了崭新的防范与保护措施。信息隐藏与数字水印技术就是一种新的数字媒体保护措施,它是将特定的信息(如版权信息、秘密消息等)嵌入到图像、语音、视频及文本等各种数字媒体中,以达到标识、注释及版权保护等目的,同时,这种信息对宿主媒体的影响不足以引起人们的注意且具有特定的恢复方法,此信息对非法接收者应该是不可见、不可察觉的。数字水印技术是保护多媒体数据版权和图像可靠性论证的一种新技术。本人基于目前国内外图像数字水印研究现状做了如下的工作: 1.为提高数字水印技术的安全性,比较研究了几种常见的图像置乱技术各自的特点以及如何用MATLAB 工具编程实现。2.在DCT 域中,通常将水印嵌入到AC 分量上,特别是在中频上。近来,有学者提出:图像经过DCT 变换后,它的直流分量的感觉容量比交流分量大。从稳健性的角度,DC 分量最适合用来嵌入水印。作者提出了一种基于DCT 域DC 分量上嵌入水印的算法,旨在用实验验证这一论点。3.学术界有人认为典型的DWT 数字水印算法的性能要比DCT 数字水印算法要好些,但目前又缺乏足够的证据来证实这一观点。作者用实验数据定量比较分析了传统的DCT 数字水印算法与DWT 域数字水印算法的性能特点。4.水印的强度决定其鲁棒性,多数水印算法是人为地调整水印的嵌入强度,其缺点是只对少数图像适用,针对这一问题提出一种在DCT域中结合纹理特征的自适应数字水印算法,适合于多数图像,实验结果证明算法是可行的。5.早期数字水印多采用伪随机序列(如M-序列)或高斯噪声作为嵌入水印,检测算法只能给出有无水印的结果,不能满足版权保护的要求,并且目前水印算法多针对灰度图像,少有的文献研究鲁棒性彩色图像多水印的嵌入与提取方法。因此,利用小波变换多分辨率特性和人眼视觉特征,提出一种新的彩色图像多水印算法。更多还原

【Abstract】 The rapid growth of the information technology, the development of e-commerce, the increasingly easy access to digital media and the increasingly powerful digital media manipulation tools have made media security a very important issue. In recent years, some new concepts of information security have been proposed, especially in the application of intellectual property right protection, copyright protection and Tamper-proofing. Information hiding and digital watermarking are new measures of digital media protection, which are processes of embedding hidden messages within digital media such as text, images, video and audio. The hidden messages could be copyright information, user distributor identification number, company logo, personal signature, control bits, or any information considered useful for a specific application. Digital watermarking has recently been proposed as a new mean to provide copyright protection of multimedia data and image authentication. Author has done following works that based on the stuied actuality of domestic and overseas in watermarking domain. 1. To improve the security of digital watermarking technology, compared and studied several familiar Image Shuffle technology, used MATLAB tool to realize it. 2. In DCT domain, watermark is usually embedded into AC components, especially into intermediate frequency domain. Lately, some scholars put forward that the capability embedded into Direct Current component is more than that of AC components. Author advance a arithmetic based on embedded watermark into DC component of DCT domain to validate this issue that DC components is fit to embed watermark. 3. At present, some academicians deem that performance of representative digital watermarking methods in DWT domain is better than in DCT domain. But the evidence to approve the viewpoint is absence. Author compared and analysed the performance and specialty更多还原