【作者】 宋林峰；

【导师】 朱灿焰；

【作者基本信息】 苏州大学 ， 通信与信息系统， 2008， 硕士

【摘要】 随着数字多媒体的广泛应用和网络技术的快速发展,对数字产品版权保护的需求日益迫切。数字水印技术为数字作品的版权保护提供了一种新的途径。它通过把版权标识隐藏在数字作品中,并能够从中提取出来以证明作品的真正所有者,从而达到保护版权的目的。但是,在数字媒体传输过程中会无意或有意受到攻击,对数字水印构成了很大威胁。为了能有效地保护数字作品,对各种攻击具有较好鲁棒性的数字水印技术的研究成为商业界和学术界共同关注的热点。本文的工作和贡献主要体现在以下几个方面:首先,系统分析了基于分块奇异值分解的数字水印算法,此算法对抵抗JPEG压缩、线性及非线性滤波具有很强的鲁棒性。但缺点非常突出,主要包括高强度水印条件下出现方块效应、对噪声敏感、对剪切攻击和旋转攻击及高JPEG压缩鲁棒性差。针对这些突出问题设计了基于平稳小波的数字水印算法。算法中结合人类视觉模型推导出了含水印图像质量度量PSNR与平稳小波域中水印嵌入强度的关系式,为实现不可见性与鲁棒性的均衡和自适应地嵌入最大强度水印提供了依据。并从加强剪切、旋转攻击的鲁棒性方面考虑,对水印系统加入预处理操作。实验结果显示该算法具有优良的鲁棒性和安全性。其次,为提高水印纠错性能,将纠错编码与盲水印系统结合提出了基于turbo码的盲数字水印算法。算法中充分利用了平稳小波冗余特性及空间平移不变性、图像矩阵最大奇异值抗扰动特性和人类视觉特性,并在平稳小波水印系统基础上对嵌入算法作了适当改进。实验结果显示,此算法对线性或非线性滤波、JPEG高压缩率和噪声攻击具有很好的鲁棒性能。更多还原

【Abstract】 With the rapid development of networks and the wide application of digital multimedia technology, the copyright protection of digital products becomes increasingly important. The watermarking technology provides a new approach of digital copyright protection. By hiding the copyright identification in products and watermarks can be extracted to prove the true owner of the product, which can get the purpose of protecting the copyright. However, the watermarks may be attacked accidentally or intentionally, which threaten the security of digital watermarks seriously. In order to protect the digital product effectively, the research of watermarking technology with good robustness has already been focused in both academia and business field.The work of this paper can be concluded in the following aspects:Firstly, the digital watermark algorithm based on the block singular value decomposition (SVD) is analyzed detailedly. This algorithm has robustness in JPEG compression, liner and non-liner filter attacks. However, it has the following problems such as serious blocking effect at high embedding intensity and low robustness at the noise or the JPEG high-compression condition and so on. To solve these problems, a digital watermarking algorithm based on the stationary wavelet is proposed in this thesis. Combining the Human visual characteristics (HVS), the relation between PSNR and embedding intensity in stationary wavelet transform (SWT) domain is derived. Then the equalization between invisibility and robustness and embedding watermark with the maximum intensity adaptively is achieved. And some pretreatment is integrated into the watermarking system to enhance the robustness in cropping and rotating. The simulation shows this algorithm has good performance.Secondly, to enhance the error correction, a new blind digital watermarking algorithm based on turbo code is proposed, which integrates the error correction coding with blind watermarking. This algorithm makes full use of the redundancy and space translation invariance of stationary wavelet, the disturbance rejection of the maximum singular value and the characteristic of human visual system. And the embedding algorithm is also improved. The simulation results prove that this watermarking system has good performance of robustness against some attacks such as liner and non-liner filter, JPEG high-compression and noise, as illustrated in this thesis.更多还原