【作者】 李林国；

【导师】 顾耀林；

【作者基本信息】 江南大学 ， 计算机应用技术， 2008， 硕士

【摘要】 随着计算机图形学,游戏,虚拟现实等领域的发展,三维物体的应用也越来越广泛,随之而来的是非法占有、复制、修改、传播未授权产品的侵权行为越来越多。从而使人们对其产品的所有权、修改权等合法权利受到严重破坏。数字水印技术作为一种新的信息安全保护措施,在三维物体领域的应用已经成为一个研究热点。当前的数字水印技术主要集中在静止图像,音频,视频,电子文档等数据领域,而对于三维物体数字水印领域的研究相对较少。但是随着社会和个人对知识产权保护意识的增强,三维物体数字水印具有广泛的应用前景。本文在图像,音频,视频等数字水印研究的基础上,讨论了三维物体水印方法,并提出了空域和频域两种不同的数字水印方法。在空域方法中,本文研究了三维物体数据结构和仿射不变空间理论,将三维物体变换到仿射不变空间,利用三维物体数据源中顶点的位置关系,巧妙的实现了在球面坐标下嵌入水印。为了保证水印的不可见性,分别使用了中值和平均值两种方法,实现了三维模型水印良好的视觉效果。实验结果表明了该方法的有效性。在频域方法中,本文深入研究了离散余弦变换(DCT)和小波变换(DWT)两种频谱分析方法。相对于空域方法,频域方法在鲁棒性方面的效果更好,但是基于三维物体特殊的数据结构,寻找频域分析的数据比图像数字水印要困难的多。本文的频域方法,首先借鉴空域方法将三维物体变换到仿射不变空间,这样就可以在一定程度上保证水印的几何鲁棒性。然后将三维物体变换到球面坐标,根据球面坐标的角度值来选择嵌入点,通过改变球面半径来嵌入水印,并充分利用三维物体的数据结构,尽量保证算法的快速性。由于高维离散余弦和小波变换在计算量和计算速度上有很大的劣势,因此最后我们通过由球面半径组成的一维信号来嵌入水印,保证了水印的嵌入速度和鲁棒性。在试验结果分析中,本文通过图表的方式展示了试验结果,并对空域和频域水印进行了比较。更多还原

【Abstract】 With the development of computer graphics, games and virtual reality etc. The 3D objects are applied widely. And then the danger of illegal owned, copying, changing, transmitting copyrighted data without authorization is generated. The owners’copyrights are treated roughly. As a useful information security technique, the 3D object watermarking is becoming one of the research interest in the field of watermarking.Current watermarking technology focuses on media types like still images, audio and video streams, digital documents and so on. In contrast, the problem of watermarking 3D objects has received less attentions from researchers. But following the consciousness of copyright protection, the community or the individuals are paying more attention to it. So the 3D objects watermarking will possess extensive application perspective in this area.In this article, we have analyzed the technologies of 3D objects watermarking based on the watermarking method of still images, audio and video streams. And presented two methods respectively in the spatial domain and transform domain.In spatial domain methods, we analyzed the data structure of 3D objects and affine invariant space. First, a transformation of the model to an affine invariant space is finished prior to watermark embedding. Using the relationship of 3D objects’original data ,embedding the watermarking in the radius of spherical coordinates. Considering the visual masking, we used median and average method to embed the watermarking, and experiments show the availability of the result.In transform methods, we researched the DCT(Discrete Cosine Transform) and DWT(Discrete Wavelet Transform)deeply. Contrasted the spatial domain methods, the transform domain methods have better robust result. But based on the special data structure of 3D objects, it is more difficult to find the transformed data than the still images. In order to using transform domain method, we transform the objects to an affine invariant space as the spatial domain method,and it is useful to the geometry robust in some degree. then we change the 3D object to the spherical coordinates, choosing the embedding positions by the angle of spherical coordinates, and embed the watermarks in the radius of spherical coordinates. Because the slow speed and complicated of high dimension DCT and DWT, we produce a 1D signal from the radius of spherical coordinates, and perform a transform on it, then embed the watermark by changing the frequency coefficients. In experiments analysis, we show our result through images and tables, and compared the spatial method with the transform domain method.更多还原