【作者】 王索真；

【导师】 杨兴强；

【作者基本信息】 山东大学 ， 计算机应用技术， 2008， 硕士

【摘要】 数字化技术的发展为信息的获取和传播提供了极大的方便,同时也带来了数字化产品的版权保护问题。数字水印技术为数字化产品的版权保护提供了一种有效方法。目前的数字水印大多是针对图像,音频和视频,而对三维模型水印技术的研究工作相对较少。但是,基于CAD的三维模型的应用越来越多,从在网络中提交或出售三维模型产品的公司或版权所有者的立场来看,他们同样面临版权保护问题,为了保护三维模型在网络上共享而不被非法拷贝,迫切需要一种能够防止未经授权而非法使用已发布的三维模型的方法,数字水印技术作为一种有效的版权保护手段在这个领域具有广泛的应用前景。Ohbuchi在1997年首先提出三维模型的数字水印技术,之后虽然得到一定的研究,但是由于三维模型数据的特殊性,比如缺乏像图像等数据的自然的频域分解算法,这一领域的数字水印技术还不成熟,本文在参考国内外三维模型数字水印技术研究现状的基础上,主要研究三维网格模型的数字水印技术。本文首先介绍了三维模型数字水印技术的研究背景和意义,以及数字水印的一些基本概念,重点阐述了三维模型数字水印技术的特性和主要分类,简单介绍了水印技术的相关工作和研究现状。根据水印嵌入技术的不同,可以将其分为空域三维模型水印技术和频域三维模型水印技术,本文按照这种分类方法对现有的一些典型的三维模型数字水印算法进行了归纳总结和分析,分别指出了各算法的优势和不足之处。基于三维网格模型的特性和水印系统的功能分析,本文设计并实现了一个完整的三维网格模型数字水印系统,介绍了三维数据格式的组织与存储方式,简练而又清晰地说明了系统框架中各模块的实现,包括水印的生成和检测,集成了所有常见的针对三维模型的攻击操作,比如仿射变换、网格简化、噪声、裁剪、网格光顺攻击等,对各种攻击算法的实现进行了简单介绍。对于一些攻击来说,在进行水印提取之前可能需要进行预处理,即网格的重定位和重采样,由此可提高水印提取的正确度,因此对于整个水印过程非常重要,文中详细阐述了一种典型的重定位和重采样的算法。本文参考图像数字水印技术的研究方法,提出了一种基于DCT中频的三维网格模型数字水印算法,将代表三维模型几何特征的“半径”距离看成一维的数字信号,使用DCT变换到频域后在中频系数上加入数字水印,再通过逆向DCT转回空域。实验结果表明该算法实现简单,在保证水印透明性的前提下对于简化攻击、裁剪、随机噪声攻击等具有良好的鲁棒性。E Praun提出的基于渐进网格的三维模型数字水印算法的鲁棒性很高,但其所采用的Hoppe的渐进网格的构造方法十分复杂,影响了总体水印算法的效率,本文详细介绍了渐进网格的原理,并给出一种简单实用的网格塌陷方法,提高了渐进网格的构造效率同时保证渐进网格的有效性,实验结果证明,基于它的水印算法同样具有良好的鲁棒性。本文最后总结全文,并给出了对三维几何模型数字水印技术将来研究方向的讨论。更多还原

【Abstract】 Digital technology facilitates the distribution and storage of information, but brings out the problem of copyright protection. Digital watermarking is one effective way to solve the problem. Most of the current watermarking technology focuses on media types like still images, audio and video. In contrast, 3D watermarking has received less attention from researchers. But, more and more applications depend on CAD-based 3D data, accordingly, companies or copyright owners who present or sell their products on internet will face copyright protection problem. The straightforward demand is to prevent their 3D-based material from unauthorized use. As a new way for the protection of information copyright, digital watermarking possesses extensive application perspective in this area.First watermarking algorithm of 3D model is brought by Ohbuchi, and later some research had been done. But watermarking of 3D model technology is not mature because 3D model data is distinguishingly, especially 3D model can not be decomposed like image. We focus on research of 3D mesh watermarking based on the research state at home and abroad.We first introduce the background and meaning of 3D watermarking technology, and some basic concepts of digital watermarking. Then we expound the characteristic and classification of digital watermarking of 3D models. Then, we describe briefly the related work and research. According to the difference of watermark embedding approaches, three dimensional model watermarking can be divided into two kinds: Embed watermark in spatial domain and in frequency domain. According to this classification, we conclude and analyze the classic digital watermarking algorithms of 3D models, and describe their advantages and disadvantages.We design and implement a complete three-dimensional model watermarking system based on the characteristic of 3D model and function analysis of watermarking system. We introduce the organization and storage of 3D model data, and explain the implementation of each module in the frame briefly and clearly, generation and detection of watermark involved. The system integrates all of the common mesh attack operations, such as affine transformation, mesh simplification, random noise, cropping, smoothing etc, all of these algorithms are present. For some attack operations, preprocessing is needed sometimes before extract the watermark. The pre-processing including mesh registration and re-sampling, which can raise the exactness of watermark detection, is very important to the whole watermarking process. One typical algorithm of registration and re-sampling is introduced particularly.In this article, we propose a robust 3D mesh watermarking algorithm based on DCT intermediate frequency coefficients, consulted the research method of image watermarking. First, we take the "radius" of 3D model which represents the geometric characteristic of 3D model as a digital signal in spatial domain and transform it to signal in frequency domain using one-dimensional discrete cosine transform (DCT). Then we embed watermark into the intermediate frequency coefficients of signal in frequency domain, and at last transform it back to a digital signal in spatial domain using inverse DCT. Experiments show that this algorithm is both simple to implement and robust against common mesh attacks such as mesh simplification, cropping, random noise and so on with good transparency. The 3D model watermarking algorithm based on progressive mesh which proposed by E Praun is one of the most robust algorithm, but this scheme employed Hoppe’s progressive mesh construct method, which is very complex and influences the efficiency of watermarking algorithm. We introduce the principle of progressive mesh detailedly, and give a simple and practical method for mesh collapse, which can be used in the watermarking process to improve the efficiency and guarantee the validity of progressive mesh construction. Experiments show that the watermarking algorithm based on the improved progressive mesh is robust too.In the end we summarize my works and some possible research directions of 3Dmodel watermarking in future.更多还原