【作者】 刘畅；

【导师】 刘志镜；

【作者基本信息】 西安电子科技大学 ， 计算机应用技术， 2006， 硕士

【摘要】 计算机网络和通信技术的迅猛发展,使得数字信息的产生、复制、传播更为容易。在这个背景下,如何防止具有版权的数字数据在未经允许的情况下,被非法的使用或是传播,这就提出了新的研究的课题:就是如何运用现有的理论和技术,更好的解决数字产品所有者的合法权益的保护问题。数字水印技术就是在这样的背景下诞生的,随着研究工作的不断深入,其理论体系日趋完善,应用领域也不断扩大。因此,对数字水印的研究不但有其重要的理论意义,更有着十分重要的实用的意义。本文就数字水印技术做了以下的工作。首先,对数字水印的概念、框架、算法做一个简要的介绍;其次,主要叙述了数字图像水印的特性、嵌入和提取的算法、常见的攻击;然后,简单回顾一下算法的核心理论小波理论。最后,重点提出了两种新颖的基于小波的数字图像水印的算法。一是适用于数字图像版权保护的鲁棒性水印算法。这种方法向宿主图像嵌入的信息是一幅数字图像,先将这个水印图像分割为两部分,一部分水印图像被嵌入宿主图像的小波域上,另一部分水印图像直接嵌入宿主图像的空域中。在提取水印的过程中,分两个步骤得到水印图像的两个部分,再将这两个部分合并在一起得到有意义的水印图像。对于大部分主要的攻击,这种方法表现出了良好的性质。该水印方法具有小波域上嵌入水印的鲁棒性的同时增加了嵌入图像的信息容量,是一个很好的向宿主图像中嵌入大量水印的策略。二是适用于图像认证的半脆弱水印算法。该算法通过在小波包上的奇异值分解中采用一个简单的量化索引调制(QIM)处理来进行图像认证。在原始图像的小波包子带的奇异值块中嵌入信息。为了提高嵌入水印后的图像的保真度和视觉质量并增强水印的安全性,我们建立了一个基于子带块统计的自适应的量化参数模型。该方法对于JPEG压缩具有很好的鲁棒性,同时对于恶意的一些操作如滤波、加随机噪声等攻击非常敏感,具有半易损水印的特点。水印的提取过程是盲检测的过程,在不需要原始图像的情况下只需要量化参数就能够进行检测。更多还原

【Abstract】 With the rapid development of computer network and communications, it is easier to generate, copy, transmit the digital information than ever before. In this situation, the new subject has been presented to prevent the transmission of digital data without authorization. That is digital watermarking, which embeds an imperceptible signal (the watermark) into multimedia data such as image, video, and audio, and it has attracted considerable attention in the past few years.In the thesis, we first introduce the history and origin of information hiding and digital watermark techniques, and then discuss digital watermark techniques including the conception of digital watermark, the applications of digital watermark, the classification of watermark technology and basic requirement, the composing of watermark system and general theory, and the classification of watermark attacks. At the last, we proposed two new image watermarking algorithm based on wavelet.The first scheme is robust watermarking algorithm for image copyright protection. We propose a new watermarking scheme by embedding a digital image signature/watermark in the wavelet domain and spatial domain. The watermark is disparted into two parts. One part is inserted in the low frequency sub-bands of wavelet domain. The other part is inserted in spatial domain directly. There are two steps in the watermark extraction .We can get the original watermark image from the two steps. This method not only has the robustness but also increase the capacity of hiding data.The second scheme is a semi-fragile watermarking algorithm for image authentication. We propose a novel, yet simple, image-adaptive watermarking scheme for image authentication by applying a simple quantization-index-modulation process on wavelet packet domain singular value decomposition. The proposed scheme is based on bit embedding on the singular value of the blocks within wavelet packet subband of the original image. The scheme is robust against JPEG compression but extremely sensitive to malicious manipulation such as filtering and random noising. Watermark detection is efficient and blind in the sense only the quantization parameters but not the original image are required.更多还原