基于小波变换的二值图像技术研究

【作者】 张黎；

【导师】 杨峰；

【作者基本信息】 山东师范大学 ， 计算机应用技术， 2007， 硕士

【摘要】 随着多媒体技术和计算机网络技术的发展,数字产品的版权问题显得尤为重要,如何确保数字产品的安全已经成为国内外研究的热点。数字水印技术作为一种新的版权保护方法,从20世纪90年代提出至今,已经受到广大科技工作者和用户越来越多的关注,研究成果也是层出不穷,然而针对二值图像水印技术的研究还相当有限,由于二值图像自身的特殊性.与一般的图像水印有很大不同,用于灰度图像和彩色图像的数字水印隐藏方法一般很难应用于二值图像,而在实际中以二值形式出现的数字产品还是比较广泛的,如黑白图片、文档、签字、支票、合同、证明和印章等,其中黑白图片是黑白二色,支票可能是黑黄二色,而印章可能是红白二色。因此对它们的版权保护和防伪鉴定也是数字水印技术领域的一个重要研究方向。大部分现有的二值图像数字水印技术都是基于空间域的方法,隐含的信息量小,需记录的隐藏位置信息量多,嵌入和提取过程复杂,抗攻击能力差。针对大多数数字水印嵌入过程中数字水印鲁棒性和不可感知性之间的矛盾,讨论了一种新的数字水印技术——零水印。零水印是利用数字作品自身特征来构造所要嵌入的水印,从而在嵌入水印时不必改变原数字作品的信息。本文首先对课题的背景和当前国内外研究现状做了简单的回顾,着重讨论了二值图像数字水印的相关技术,归纳了水印技术的分类,分析了空间域图像水印、频率域图像水印的优点与缺点,给出了几种常用的检查并判断水印系统合适与否的指标及相应的计算方法。针对目前二值图像中嵌入水印的算法对于几何攻击的抵御较差,本文提出了一种基于全局DWT变换的鲁棒性数字水印算法,该算法有较好的抗几何攻击和噪声攻击的能力,且是盲水印,提取水印时不需要原始二值图像。主要思路是将二值图像进行小波变换,将变换后的系数进行二值化,观察发现小波高频系数绝对值较大,做几何变换之后系数的改变基本上不影响其二值化的结果,因此利用高频系数二值化后的矩阵与置乱加密后的水印图像进行异或运算生成相应的一个逻辑表,该逻辑表作为密钥,通过向第三方注册,来保护自己对原作品的版权。同时,通过这个密钥(Secret Key)可以提取出二值图像中与之对应的水印以证明自己对数字媒体的所有权。最后通过Matlab6.1模拟各种噪声攻击和几何攻击,试验证明该算法简单易行,嵌入水印容量大,具有更好的鲁棒性和隐蔽性,该方法具有一定的理论和应用价值,为二值图像数字水印发展提供了一些新的思路。更多还原

【Abstract】 With the development of multimedia and network techniques, the Copyright Protection of digital media is becoming more and more important. How to ensure the security of digital media has become the focus of research work all over the world. As a new method of Copyright Protection, digital watermarking techniques has attracted more and more attention result are obtained. However, the technique research aiming at binary images is limited.Because of the characteristic of binary images itself,the binary images watermarking have big difference to the common watermarking,the data hiding methods which is used in gray image and color image is hard to be used in binary images.In practice, there are o lot of binary image such as black-white-photo、document、signature、contract、seal. So it’s an important research direction for their copyright protection in digital watermarking domain. Most of now available binary watermark algorithm are based spatial domain, there embedded information are limited and the operation process is complicated. For the conflict between robustness and imperceptibility during embedding, a new water- marking technique: non-watermarking is introduced. Non-watermarking utilizes the feature of the original works to construct a watermark, but does not modify.This paper first reviews background of subject and the current development condition, point out the technology of binary image watermark, sums up the types of watermark technology, analyzes the image watermark’s advantage and shortcoming of spatial and frequency domain and proposes a appropriate target of watermarking system and it’s computed method. A robustness algorithm is given which are based on DWT domain. This Algorithm has good ability to resist geometric and noise attacks. Without original image the watermarking can be extracted ,so it is a practical blind watermarking. First, Binaries the coefficient of the image that is wavelet transformed. Second, we use the high frequency part add to the watermark-image that disordered, so we obtain a logical array to protect copyright. The experiment result shows that the algorithm is simple and increased the information embedded and has better robustness and concealment. It is consequently known to us that the method brought forward in this paper has certain value of theory and application, under the wish to bring some novel thought for development of binary digital watermarking.更多还原