外文翻译及文献 基于DCT变换的水印算法实现

1、毕业论文（设计）外文翻译题目:专业:班级:学号:姓名:指导教师:基于DCT变换的水印算法实现基于子带离散余弦变换DCT应用于图像水印的技术基于子带离散余弦变换 DCT应用于图像水印的技术已经被提出并应用。水印是波在所有选定的含有若干系数的四个频带段的1级分解。应用大量的系数使每个波段给出了不同的检测输出结果。其结果是采取平均检测结果的所有频段的值。结果表明，最终的结果是优于所检测输出的每个波段所得的结果的，从而实现了非常强大的水印方案。1、导言数字媒体技术在当今社会已被大范围的使用, 从而促使其创立知识产权来保护。 就其性 质而言，数字媒体是能够 100被完整复制的，因此，必须采取有效的标识系

2、统（是显而易见的）。这就是水印的由来。水印技术是指将无法被看见的数据埋入图像中，从而确定合法 的创建者 /拥有者。水印应当具有健全的可以适用于（抵挡）各种各样的图像攻击的技术。一些常见的攻击包括过滤，压任何尝试从原始图像删除所有权信息的方法（被称为）攻击。缩,直方图修改,剪裁,旋转和缩小。 （水印）主要有两个嵌入方向,即空间域和变换域。变换域的技术对普通的图像攻击技术更敏感，如过滤或JPEG 压缩。这种转换技术通变换域技术在图像水印中是最受欢迎的。 在这种情况下， 图像技术正在通过某些常见的， 频繁发生事情改变着， 并且使得水印转换系数被高度完美的应用于图像上。常使用DCT （离散余弦变换），

3、DFT （二维傅里叶变换）和 DWT （离散沃尔什变换）。许多发生在现况下的一个问题是各种数量和位置的改变将使其在图像中频繁的变化着。有效方法已经被提出，其中大部分是源于科克斯（Cox ）的体系。皮瓦等人扩展了这一方法，从而提出了一种隐藏检测系统（ blind detection system ）。在这些全部情况中都将图像处理作为一个整体，但一些系数变化不超过16000,通常的图像尺寸是512x512。由于大多数的过程都是在数字统计的背景下（执行的） ，因此我们宁愿使用越多系数越好。这就是为什 么我们建议使用子带离散余弦变换 DCT 原因。第 2 节，我们将围绕目前的子带 DCT 分解层和模型

4、的参数进行讨论。第 3 节，我们将 所阅读到的方案进行测试, 并且解释每一个波段的处理后的情况下, 最后检测未经过处理的 五个通常攻击方法。最后，我们将在最终章节结束这一讨论。2、子带 DCT 和水印的模型郑和米特拉（ Jung and Mitra ）已经于 1996 年介绍了子带 DCT 。这是一种涉及小波变换 和离散余弦变换 （DCT ）的方法。 将原始图像二次抽样后经过高通滤波过滤器和低通滤波过 滤器（处理） 。结合这两个过滤器的各个方向（横向和纵向）的过滤使四个子带为每个层进 行分解。这种过程相当于通过低通滤波器向各个方向透进的频带能够进一步的被二次抽样和DCT 系数来进行转化。过滤，

5、使其能在另一层被分解。最后，使每一个频带都通过使用我们尝试在我们的实验中， 我们不得不选取并使用一定数量的的分解方法和小波技术。用大量的分解层，即 1，2和 3层来进行试验。实验结果表明， （分解系数 ）并没有在某一层的探测结果上有重大的提高,与此同时, 图象退化的现象却更容易被发现。 因此, 我们把已经被分解的原始图像的四个频段放入一层内。这种娴熟的使用最简单的小波技术的方法，是哈尔（Haar）提出的。下一步是将每一频段进行DCT变换。为了解释我们的水印技术，我们使用下列公式:tj =ti +ati Xi其中ti是正在转化的系数，ti是水印系数，xi是一个被用于水印中的随机序列的高斯分布。参

6、数a是与模型的浓度有关。我们对其使用两种不同的准则，一种是LL-频段，剩下的将使用另一种频段，那就是当a=0.1的时候使用LL-频段，a=0.2时则用另一种。这样做的理由是，低频段更容易变化，也就是说，一些细微的变化更加的明显。i参数的范围是从1到20000,使其在一个令人满意的80000系数间的变化。在每个频段中，我们都将从以5000为系数的锯齿形依次扫描。基于块分类和DCT域的图像水印算法摘要：本文提出了一种基于离散余弦变换（ DCT ）域图像水印算法。图像水印算法有两个阶段： 特征嵌入和特征检测。 第一阶段，它将一个标识符号嵌入进图像。第二阶段是被公认（已知）的。该算法有两个处理步骤。第

7、一步无疑是选择像素区块并使用参数进行设置，而第二个步骤是将DCT系数强制的嵌入在选定的区块内。两种不同的参数规则表明修改 DCT参数系数出现频率的重要性。第一种方法是将DCT规则嵌入到选定的线性约束内，而第二种方法则是按照所给予的特定参数进行循环检测。上述所提到的水印算法是不能在 JGEG压缩和过滤条件下使用的。1、导言数字水印是当今广播电视和密码技术的探讨的大体背景下产生的。为了避免（他人）未经授权就发布图片或其他多媒体资源，已经提出了大量的解决方法。其中多数是提议做一些难以被发现的图片修改以供以后使用。这种图片修改技术被称为水印。水印是将图片做一些不明显的修改（以确定版权所有），从而能够强

8、力的抵制可能出现的各种图像处理技术。水印算法已经被大量的颁布过。 他们不是随机性的就是确定性的。 这些算法， 包含了图 像强度域和变换域。在中频范围内 DCT 的变换系数受嵌入的 8*8 像素块所约束。将授权信 息嵌入 DCT 系数后应用所得的 DCT 系数来处理整个图像。图像水印算法有两个阶段： 特征嵌入和特征检测。 通过特征嵌入来编写绝对代码分配给 所有者后让其嵌入图像。 在检测阶段用算法来确定所规定的代码。 信号检测理论是一种对许 多领域都有效的应用技术。 水印图像能够用许多不同的处理方法来转变图像和处理运算法则 来防止其被摧毁， 这就是数字水印技术。 图像压缩是每个图像都有可能经历的图

9、像变化过程。标准的静态图像压缩算法是JPEG格式。JPEG格式是基于尽可能减少资源在离散余弦变换DCT ）域上的消耗（而产生的） 。受损压缩，即使图像的信息遭受损失的压缩方法是在高 频域上发生的。在拟定的水印算法中，图像被分割成类似JPEG算法的8*8像素块。该水印算法包括两DCT 参DCT 系数上使其个步骤。第一步是依据高斯网来选择某些特定的块。在选定的块中，我们通过修改 数来使其强制完成某一给予的约束。该参数是把高斯函数加在系统规定的 组成水印代码。在检测阶段我们首先检测 DCT 参数，然后检测各自的区块的位置来确定是 否被篡改。A SUBBAND DCT APPROACH TO IMAG

10、E WATERMARKINGA subband-DCT approach for image watermarking is proposed in this communication. The watermark is casted in a selected number of coefficients of all four bands of a one-level decomposition. A great number of coefficients is being used. Each band gives a different detection output. The

11、result is taken as the average detection result of all bands. It is shown that the final result is better than the detection output of each individual band, thus leading to a very robust watermarking scheme.1. INTRODUCTIONThe great spread of digital media in nowadays, has urged for the protection of

12、 the intellectual property rights of the creators. By their nature, digital media are 100% reliably copied, so the need for an effective marking system is obvious. This is where watermarking comes in. Watermarking stands for the embedding of perceptually invisible information into image data that id

13、entify the rightful creator/owner. Watermarks should be robust to various image attacks. Every attempt to remove the ownership information from the original image is called an attack. Some common attacks include filtering, JPEG compression, histogram modification, cropping, rotation and downscaling.

14、 There are two main directions for embedding, namely the spatial and the frequency domain. The spatial domain techniques are more vulnerable in common image attacks such as filtering or JPEG compression.The frequency-domain approaches are the most popular for image watermarking. In these schemes, th

15、e image is being transformed via some common frequency transform and watermarking is achieved by altering the transform coefficients of the image. The transforms that are usually used are the DCT, DFT and the DWT. A question that occurs in such approaches is the number and the position of the altere

16、d coefficients in the frequency representation of the image.s et al system. PiMany different ideas have been presented, most of them originating from Cox et al have extended this idea, thus providing a blind detection system. In all these cases the image is being processed as a whole but the number

17、of coefficients altered is not more than 16000, the usual image size being 512x512. Since most of these processes have a statistical background, we would prefer to use as many coefficients as possible. This is why we suggest the use of subbandDCT.In section 2, we present the subband DCT, the decompo

18、sition levels and discuss the casting scheme and parameters. In section 3, we test the reading scheme and examine each band s individual contribution in the case of no processing and also the final detection results for five common attacks. We end with the final conclusions in section2. SUBBAND DCT

19、AND W ATERMARK CASTINGJung and Mitra have introduced the subband DCT in 1996. It is a method that involves both wavelets and the Discrete Cosine Transform (DCT). The original image is sub sampled and filtered with the use of a high pass and a low pass filter. The combination of the two filters for e

20、ach direct ion (horiz on tai and vertical) of filteri ng gives four subba nds for each level of deco mp ositi on. The band that corres ponds to low p ass filteri ng in both direct ions (LL band) can be further subsa mpled and filtered thus p rovidi ng ano ther level of deco mp ositi on. Fin ally, ea

21、ch of the bands is tran sformed with the use of the DCT.For our exp erime nts, we had to select the nu mber of deco mpo siti on levels and the wavelet to be used. We tried differe nt nu mbers of deco mpo siti on levels, n amely 1, 2 and 3. Exp erime ntalresults have show n that there was nny sig ni

22、ficarat impro veme nt in the detect ion results for more levels tha n one, while at the same time, the image degradati on was more easily observed. So we deco mp osed the origi nal image into one level with four ban ds. This was acco mp lished using the simpi est wavelet, that is Haar. The next ste

23、p was to p erform a DCT on each of the ban ds. To cast our watermark we used the follow ing formula:ti=ti +atiXiwhere ti are the transformed coefficients, t i are the watermarked coefficients and xis a random sequenee of Gaussian distribution, used as the watermark. The a- parameter has to do with t

24、he stre ngth of the casti ng. We use two differe nt values for it, one for the LL-ba nd and a differe nt one for all other ban ds, that is a=0.1 for the LL- band and 0.2 for the others. The reas on for this is that the low frequency band is more vulnerable to changes, meaning that slight changes are

25、 easily noticeable. The i parameter ranges from 1 to 20000, thus leading in a satisfactory number of 80000 coefficients that are altered. In each band we start from coefficient 5000 in the zig-zag sca nning order.Image watermarki ng using block site selecti on and DCT doma incon stra intsAbstract: I

26、n this paper we prop ose an image watermark ing algorithm based on con stra ints in the Discrete Cosi ne Tran sform (DCT) doma in.An image watermarking algorithm has two stages: signature casting (embedding) and signature detection. In the first stage it embeds an identifying label in the image. Thi

27、s is recognized in the second stage. The proposed algorithm has two processing steps. In the first step certain pixel blocks are selected using a set of parameters while in the second step a DCT coefficient constraint is embedded in the selected blocks. Two different constraint rules are suggested f

28、or the parametric modification of the DCT frequency coefficients. The first one embeds a linear constraint among certain selected DCT coefficients and the second defines circular detection regions according to the given parameters. The watermarks cast by the proposed algorithm are resistant to JPEG

29、compression and filtering.1. IntroductionDigital watermarks in the general context of TV broadcasting and cryptology were discussed in.To avoid the unauthorized distribution of images or other multimedia property, various solutions have been proposed. Most of them make unobservable modifications to

30、images, that can be detected afterwards . Such image changes are called watermarks. The watermark should not altervisibly the image and it should be robust to alterationswhich may be caused by various image processing techniques.Algorithms proposed for watermarking have been reported in various pape

31、rs. They are either stochastic or deterministic. These algorithms are either in image intensity domain or in frequency domain. In the middle range DCT frequency coefficients from the 8*8 pixel blocks are used for embedding a constraint. In the signature is embedded in the DCT coefficients obtained a

32、fter applying the DCT transform in the entire image.An watermarking algorithm has two stages: watermarking casting and detection. By means of watermark casting a specific code assigned to the owner is embedded in the image. In the detection stage the algorithm identifies the given code. Signal detec

33、tion theory is a well-established field with many applications. A watermarked image can be processed by means of various image transformations and processing algorithms which may be able to destroy, intentionally or not, the digital watermark. Image compression is the most likely transformation that

34、 an image may undergo. The standard still image compression algorithm is JPEG. JPEG is based on the minimization of the energy in the Discrete Cosine Transform (DCT) domain. In the case of lossy compression, the image suffers information loss in the high frequency domain.In the prop osed watermark c

35、asti ng algorithm from, the image is p artiti oned in 8*8 pi xel blocks similar to the JPEG algorithm. The watermarking algorithm consists of two steps. The first step selects certain blocks according to a Gaussian network. In the selected blocks we modify DCT coefficients such that they fulfill a given constraint. The parameters of the Gaussian functions and of the imposed constraints on the DCT coef