随着互联网和多媒体技术的飞速发展，文档的版权保护和防篡改问题变得越来越重要，文档水印技术是解决版权保护和防篡改问题的一种有效方法。本文介绍了文档水印的起源、应用、特征及分类，简述了常见的几种文档格式及其特点，重点研究了可编辑文档的保护技术，提出了基于新造字的文档水印算法，构建了一种双水印系统。 　　基于新造字的文档水印算法根据所有者的相关信息，创建具有一定意义的TrueType新字符，将新造字字符嵌入到文档中，利用原始文档中的字符覆盖新造字字符，在接收端通过水印提取程序进行所有权的认定。该算法适用于DOC、RTF、HTML等格式文档，能解决文档的所有权认证问题。仿真实验表明算法不可见性好，抗攻击性高。 　　在新造字和字符颜色特征两种算法基础上，作者构建了一种双水印系统。字符颜色特征的水印算法利用冗余信息进行水印定位，可在一定范围内识别篡改区域，能用于文档内容的真实性认证。双水印系统则利用两种算法各自的特点分别解决数字文档的版权保护和防篡改问题。仿真实验表明：双水印系统具备所有权认证和篡改定位两种功能，较好地提高了水印的抗攻击性。

With the rapid development of Internet and multimedia technologies, the protection and the prevention of tampering of document copyright have been increasingly significant. Document watermarking technology is an effective way to solve the problem. The origin, applications and features of document watermarking are described, and so does the classification of the existing document watermarking algorithm. The characteristics and formats of common document are also outlined. This thesis mainly investigates the protecting technique of editable document. It proposes a document watermarking algorithm based on new characters, and constructs a bi-watermark system. 　　Based on the relative information of the owner, the algorithm creates some TrueType new characters that have certain meanings. Then, the new characters would be embedded in the document, and are covered by the characters in original document. This algorithm is suitable for the document in the format of DOC, RTF, HTML, etc. It can solve the authentication problem of the document’s ownership. Simulations demonstrate that the algorithm is imperceptible and hard to destroy. 　　With the foundation of the two algorithms, new character and font color, the author constructs a bi-watermark system. The font color –based watermarking algorithm uses the redundant information to locate the watermark. It can identify the tampered region in certain area. Thus, it can be used in the authentication of the content. By using the two algorithms’ characteristics, the bi-watermark system can settle the copyright protection and anti-tamper problem respectively. Results show that the system has the two functions mentioned ahead, and the anti-attack ability is well improved.