【作者】 佘祥顺；

【导师】 王欣；

【作者基本信息】 山东大学 ， 信号与信息处理， 2010， 硕士

【摘要】 随着当今社会计算机网络技术的迅猛发展,电子产品越来越多的出现在人们的生活当中。这一方面给人们的生活带来了极大的方便,但随之而来的也带来了如何保护知识产权的问题。由于数字产品极易被篡改和复制,这给数字产品的安全性带来了极大的威胁。而数字水印的出现可以说是为保护知识产权提供了一个很有效的解决办法。数字水印技术,就是在不影响产品内容的前提下,在载体中嵌入不易被察觉的信息,从而达到保护数字产品的目的。目前许多公司已经成功开发了数字签名系统并且已经将其应用在电子商务活动中。同时数字水印技术也广泛应用在数字产品电子交易系统、票据防伪系统和影视处理系统中.数字水印技术在隐蔽通信领域中也具有非常重要的用途。在医学、制图、数字成像,数字图像监控,多媒体索引和基于内容的检索等领域,数字水印是被当做隐藏标识来使用的。与以往相比,现在的数字水印主要是在变换域中进行的,主要包括了,小波变换,余弦变换,有限Radon变换、脊波变换域水印等等。以前的研究主要是集中在小波变换域来进行的。但是由于小波只适合于表征零维(点)奇异性,而图像中广泛存在一维奇异性,即线性奇异的,例如光滑的曲线构成的物体边缘。为了更有效表示图像的特征,我们应该探寻新的方法,由此人们提出了有限脊波变换方法。自从有限脊波变换被提出以来,其就逐渐的成为了人们研究的热点,各种脊波域水印也相继被提出。本文研究的是有限脊波域的数字水印。首先详细的介绍了数字水印技术和有限脊波变换理论。然后我们将介绍集中脊波域的数字水印。我们详细的介绍了这三种数字水印方法的每一个步骤：水印的生成、水印的嵌入、水印的检测和实验结果。第一种方法是基于能量最大方向的数字水印,该方法就是要将数字水印嵌入到能量最大方向的系数当中,这里嵌入的是中频系数。第二种方法基于最大后验概率准则的数字水印方法,其基本思路就是要根据变换后的有限脊波系数的分布特点来进行估计。观察发现,变换后系数的特点符合拉普拉斯分布。我们就根据这一特性,使用最大后验准则来估计,最终得到适合嵌入水印点的系数。第三种方法是根据奇异性来选择嵌入水印的点的,我们知道小波在处理点奇异时是最优的,而Radon变换在处理线性奇异的点时是最有效的,因此我们可以通过有限Radon变换,将原始图像的线的奇异性转化为Radon域点的奇异性。再根据奇异性来嵌入水印。三种方法的效果均比较理想。本文的最后就是总结和对未来的展望,主要是对前面所提出的三种方法进行总结,并找出其中的不足然后对未来的研究方向进行安排。更多还原

【Abstract】 Now, as the rapid development of computer and network, there are more and more electronics around us. It brought great convenience, but also a problem, how to protect intellectual property rights. As digital products can easily be altered and replicated, it is very dangerous. We can say that the watermarking is a very effective method to protect intellectual property rights. Digital watermark is a new and effective way to protect digital products and to maintain the data security.It is a very practical information hiding technique. Watermarks can testify the ownership of the above. What more,watermarks can be the evidence in front of the court. In a word. Digital watermark can be an effective method to protect knowledge property and multimedia data.Digital watermarking is mainly carried out in the transform domain, including the, DWT, DCT, FRAT, FRIT and so on. Previous studies mainly concentrated on the wavelet transform domain. However, wavelet is suitable for zero degree(point singularities, Images, with its contours and borders have 1-D singularities, which wavelet is not a optimal method to process. So, we should find some new and effective method. Since the FRIT is proposed, it becomes the focus.This paper also studies the watermarking based on FRIT. First, we introduce the watermark and the FRIT. Then we detail introduce every steps:watermark generation, watermark embedding, watermark detection and experimental results.The first watermarking is based on the maximum energy direction, and our purpose is embedding coefficients to these direction. The second method is based on the MAP(Maximum A Posterior)criterion. The distribution of the FRIT coefficients is Laplace distribution.from these we choose the coefficients that embeds watermarking. The third method is based on the singularity. This method selects the coefficients by the singularity. It is more excellent for FRAT (finite Radon transform) to deal with linear singular image. However, since edges in images are typically curved, in order to obtain an effective image representation through the FRAT, the image is partitioned into blocks of side-length such that an edge appears as a straight line. By block FRAT, linear singularity is transformed into point singularity. Using the multi-scale wavelet transform to calculate the singularity of each point. Finally, the FRAT is performed for each block, the most singular coefficients are selected, and eventually the watermark is embedded.At last we summarize these three methods and propose the study direction for future.更多还原