【作者】 康守权；

【导师】 江荣安；

【作者基本信息】 大连理工大学 ， 计算机应用技术， 2008， 硕士

【摘要】 随着多媒体技术和网络技术的飞速发展和广泛应用,对图像、音频、视频等内容的知识产权保护成为迫切需要解决的问题。加密和拷贝保护机制不能完全解决这一问题:加密只在传输过程中保护数据,而拷贝保护机制又易被破坏。于是一种新的保护途径应运而生,即数字水印技术。它甚至被认为是知识产权保护的最后一道防线。数字水印技术不仅可用于知识产权的保护,还可用于认证、注释等。目前对图像等方面的数字水印方法很多且比较有效,但这些方法一般不适合应用于文本。文本数字水印技术是一个涉及密码学、图形学、视觉科学、中文信息处理、图像处理、通信及信息安全等学科的交叉边缘科学,目前在理论上是很新的一个研究方向。已经取得的文本数字水印研究成果主要集中在基于文本格式的算法方面。最为典型的有字移、行移及特征编码等。但是这类算法的水印信息不是加载到文本内容之中,因此难以抵抗打印、格式转换等常见文档处理操作。而将水印嵌入文本内容的基于自然语言处理的文本水印算法中计算机自然语言处理技术是个瓶颈。另外,一篇文档里含有多个作者版权水印信息的时候,真正的版权信息难以确定,一些论文针对这种攻击提出了很多协议,但是这些协议过于复杂,或者在有效性上执行的不是很好。本文提出并实现了一种基于字符属性的多组文本数字水印算法。WORD文档中的字符存在某些属性,其默认值的修改具有很强的隐蔽性,如字符作为WORD中的range对象,其NoProofing属性默认值保持不变而仅在编程时方可修改此值;字符的LanguageIDFarEast属性,可将其默认值改为wwdTraditionalChinese或wdChineseSingapore等。通过对以汉字为对象的特定属性值的修改可以达到嵌入水印信息的目的。实验证明,此水印算法一方面水印隐蔽性强,水印容量不受限制,另一方面,该方法具备较强的抗攻击能力并兼具较强的文本完整性检测性能。更多还原

【Abstract】 With the explosive growth in multimedia and network technologies, intellectual propertyprotection of images, audio, video etc are more urgent than ever. Encryption and copyprotection mechanisms can not fully solve the issue. Encryption usually protects data only on the transport channel and copy protection mechanisms are often circumvented easily. In this context, digital watermarking has been proposed as the last line of defense in the protection of intellectual property. Moreover, watermarking can be applied to authentication, captioning etc Now there are many effective digital watermarking methods for image etc, but these methods are not usually suitable for text.The technology of the text digital watermark is a cross marginal subject which relates to researches of cryptography, graphics, science of vision, Chinese information processing, image processing, communications, and information security, etc and is a very new research direction in theory at present. So far the acquired achievements on text digital watermarking are mainly focused on the algorithms based on text format. The most typical representatives are algorithms on word-shifting, line-shifting and character-coding. As the watermark information of this kind of algorithm is not inserted into the text content, the algorithms can not resist some common document operations such as format conversion, print processing.A multiple text calculating method based on the characters’ attributes and the words’ content is proposed in this paper, and its merits are also analyzed. It is of good imperceptibility to modify the default numerical values of characters’ specific attributes in a Word Document .For example, the character is regarded as the object of range in the Word Document, the default numerical value of NoProofing’s attribute can only be modified during programming; and the default numerical value of LanguageIDFarEast, one of the character’s attributes, can be modified to wwdTraditionalChinese or wdChineseSingapore.To embed text watermarks, we can modify the default numerical values ofcharacters’ specific attributes. Experiments show that the watermarks can be hided effectively and have unlimited capacity. Moreover, it has a good anti-attacking ability and good performance in text integrity detection.更多还原