节点文献

基于版权保护的图像数字水印研究

Research on Image Digital Watermarking Based on Copyright Protection

【作者】 许文丽

【导师】 王育民;

【作者基本信息】 西安电子科技大学 , 密码学, 2007, 博士

【摘要】 数字水印作为信息隐藏技术研究领域的重要分支,是实现版权保护或跟踪侵权行为的有效方法。它是通过一定的算法将某些秘密信息—水印嵌入到被保护的数字多媒体(如静止图像、音频、视频等)内容中,但不影响原内容的价值和使用,并且不能被人的感知系统察觉。数字水印必须难以被清除和破坏,这表明,对版权保护应用领域来说,数字水印技术必须具有较强的透明性、安全性和鲁棒性,在这个前提下,水印容量越大越好。本文主要对基于版权保护的图像数字水印及其零知识证明进行了研究,取得的主要成果如下:1.研究了数字水印系统的基本理论,包括数字水印生成技术、数字水印嵌入技术和数字水印提取、检测技术等。进一步研究了乘法水印的基于变换域系数分布模型所构造的鲁棒最优检测器及其检测阈值的确定方法,从而使得检测性能最优,或使检测错误率最低。2.结合伪随机置乱技术,提出了基于差错控制编码的数字水印方案。数字水印系统实际上可看作是一个通信系统,为了减少水印数据在载体信道中的传输错误,提高其安全性和隐蔽性,很有必要采用差错控制编码技术。采用伪随机序列对原始水印图像进行置乱,提高了水印的安全性、隐蔽性和抗剪切攻击的鲁棒性。3.提出了基于混沌置乱和混合变换域的扩频数字水印方案,该方案综合利用扩频技术所具有的抗干扰、抗衰落、抗多径干扰的能力以及很好的隐蔽性,利用混沌置乱技术能够增强数字水印的安全性和抗剪切攻击的能力以及利用混合变换域中DWT的多分辨率特性和DCT的能量压缩、去相关能力。实验结果表明,应用该方案,数字水印的安全性和隐蔽性有了很大提高,而且对剪切、平移、旋转等各种几何失真攻击和常见的信号处理(如添加噪音、JPEG压缩和滤波)攻击具有很强的鲁棒性,即使嵌有水印的图像遭受攻击后受损很严重,水印信息还是能够完好地提取出来。4.提出了基于Turbo码和CDMA的数字水印方案。CDMA是通信系统中一种有效的通信方式,具有抗噪声、抗干扰能力强,隐蔽性好,可多址复用,容量大等优点。将其应用于水印系统中,既能提高了水印的安全性和鲁棒性,还能够增加水印容量。在此基础上,为了减小水印提取过程中多址干扰(MAI)的影响,我们还采用了性能较好的编译码方法(如Turbo码)来降低接收机的检测门限,编码性能的改善直接提高了CDMA系统容量。5.对于CDMA数字水印系统,如果在一定数量(如64×64)的载体变换域系数中嵌入的水印数据量很大(如256×256),就会出现较强的同址分组干扰、宿主图像降质较严重和多址干扰等问题。对此,我们采取了三个有效措施:首先,在水印分组过程中,采用排序法构造水印分组矩阵,提取水印时,变害为利,使干扰信号变成彼此的增强信号,提高了提取出的水印质量;第二,通过对合路器合成后的编码水印序列中的每一个元素都加或减一个常数,使水印序列的幅值控制在一个适当的范围内,这样就有效解决了PSNR值下降较大的问题;第三,通过应用线性最小均方误差估计的方法对水印信息进行检测,进一步减小了MAI的影响。实验结果表明,所提出的方案和方法,切实实现了水印容量很大,宿主图像失真很小,水印的安全隐蔽性好,抗干扰、噪声以及常见图像处理攻击鲁棒性强等优点。6.在安全承诺方案和RSA公钥密码体制的基础上,提出了基于版权保护的零知识数字水印认证协议。协议采用的水印系统是乘法水印嵌入规则及其基于变换域系数分布模型构造的鲁棒优化检测器。我们还对前面提出的CDMA数字水印方案,提出了基于CDMA的零知识水印认证协议。对协议进行安全性分析表明,应用上述协议证明水印存在的同时,没有将水印、水印嵌入位置、提取密钥等敏感信息泄露出来,蓄意攻击者也就无法对水印进行修改、破坏或去除掉,协议满足完备性、正确性和零知识性。

【Abstract】 Digital watermarking is an important branch of information hiding.It is an effective method for protecting copyright and tracing piracy.Essentially,watermarking is defined as the process of embedding secret information directly into digital multimedias(such as stationary image,audio,or video signal.) and the secret information can not be perceptual by human perception system.Moreover,it is difficult to remove or spoil the digital watermarking.In copyright protection application field,there are three requirements to digital watermarking system: imperceptibility of the watermark within the host signal,the security of the watermark against attacks and robustness to signal processing.On the premise of those,the watermarking capacity should be as high as possible.In this thesis,the research on image digital watermarking based on copyright protection and its zero-knowledge proofs are performed.The main contributions are summarized as follows.1.To do researches on the basic theory of digital watermarking system, including digital watermarking generating,embedding and extracting techniques.To maximize the detection probability or to minimize the probability of detection errors,a further research on multiplicative embedding rule,the corresponding robust optimum detections and the method for obtaining a thresholds under a given false alarm rate based on the distribution models of the coefficients of transform domains are performed.2.Combining pseudorandom scramble techniques,a digital watermarking scheme based on error control code is proposed.In fact,digital watermarking system can be looked as a communication system.In order to decrease the watermarking transmission errors and to improve its security and transparence,it is necessary to adopt error control code techniques,The original watermarking image can be firstly scrambled by using pseudorandom sequence.Thus,the performances of the security, the transparency and the resistance to cropping of the watermarking are improved greatly.3.A spread spectrum digital watermarking scheme based on chaotic scramble and hybrid transform domains is presented.The spread spectrum technique has many advantages,such as the resistance to interference,signal attenuation and the multiple access interference.The chaotic scramble techniques can improve the security and the robustness against cropping.DWT has multi-resolution characteristics and DCT has the capability of compressing energy and de-correlation.It is shown from the experimental results that by using the proposed scheme,the security and the transparency of the digital watermarking are improved greatly.And the robust to geometric distortion(cropping,translation,rotation) and the common signal operations (adding noise,JPEG compression and filters) is enhanced largely.Even the quality of the watermarked image is degraded severely the watermarking can be extracted very well.4.Turbo code and CDMA basis digital watermarking scheme is presented. CDMA is an efficient communication means,which has the merits such as resistance to noise and interference,the good transparence,multiple access,high capacity and so on.When it is applied to watermarking system,the performance of the security,robust and capacity of the watermarking is enhanced greatly.At the same time,for the purpose of decreasing the multiple access interference,a good encoding method(such as Turbo code) is adopted to lower the detection threshold.Moreover,the improvement on the encoding performance enhances the capacity of the CDMA system directly.5.For a CDMA digital watermarking system,if you want to embed a large number of digital watermarking data into a certain amount of coefficients selected from transform domains,some problems will occur.Here three effective measures are adopted.Firstly,the watermarking groups are constructed by using descending sort or ascending sort.In the process of extracting watermarking,the interference signals can strengthen the being extracted signal.As a result,the quality of the extracted watermarking can be improved greatly.Secondly,the CDMA encoding sequence is added or subtracted an integer before it is embedded into the initial host image.So,the PSNR of the watermarked image can be increased.Thirdly,the influence of the MAI can be decreased by using linear minimum mean square error estimation.The experimental results show that the watermarking system do has a high watermarking capacity.And the watermarked image has too little distortion to find by human visual system.At the same time,the watermarking has a good robust performance against various attacks.6.Under the foundation of secure commitment and RSA public key mechanism, zero-knowledge watermarking verification protocols are proposed.The first two protocols adopt multiplicative watermark embedding rule and the corresponding robust optimum detection based on the distribution models of the transform coefficients.On the other hand,a CDMA basis zero-knowledge digital watermarking verification protocol is presented.All the proposed protocols ensure that a prover soundly convinces a verifier of the existence of a watermarking in a certain stego-data without revealing any knowledge(including the watermarking information,watermarking embedding position and watermarking extracting keys) that a malicious attacker can use to modify,spoil or remove the watermarking.It is shown from the security analysis that the proposed protocols do have the qualities:completeness,soundness, zero-knowledge,robustness and applicability.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络