基于几何特征的3维网格模型零水印算法  

Zero-watermarking Scheme for 3D Meshes Based on Geometric Property

作　　者：徐涛[1] 张艳宁[1] 孙瑾秋[1] 郑江滨[1] 林增刚[1] 

机构地区：[1]西北工业大学计算机学院,西安710072

出　　处：《中国图象图形学报》2009年第9期1819-1824,共6页Journal of Image and Graphics

基　　金：国家自然科学基金项目(60472072);国家高技术研究发展计划(863)项目(2006AA01Z324);陕西省自然科学基金项目(F0206);教育部"新世纪优秀人才计划"项目(NCET-05-0866)

摘　　要：由于3维网格模型对水印嵌入强度高度敏感,为避免外观视觉效果严重失真,如今多数水印算法只能采用低强度的嵌入策略,从而导致其鲁棒性能下降。为了解决水印鲁棒性和透明性之间的矛盾问题,提出了一种不对网格模型数据进行任何改动的零水印算法。该方法首先对网格模型的几何数据进行校准预处理,然后对网格按顶点包围盒进行分割来提取几何数据空域特征,进而再应用3维DCT变换在变换域中进一步集中能量,其水印由若干DCT大值系数构造生成,并在IPR数据库注册获得版权保护,该水印的检测过程无需原始网格模型数据的参与。实验结果表明,该算法构造的水印不仅能抵抗顶点重排序、平移、旋转、各向一致缩放等攻击,而且对加噪、网格简化、网格滤波和二次水印等也表现出一定的鲁棒性。3D mesh models are high sensitive to the magnitude of their embedding watermarks, and most watermarking algorithms only use low-magnitude embedding method to avoid serious distortion of visual-effect, but unfortunately,it would degrade robustness. In order to compromise the robustness and the transparency of watermarked 3D meshes, a zero- watermarking algorithm that modified nothing of 3D mesh data is presented. After calibrating mesh geometric data, mesh vertices are divided into some blocks according to their bounding-box, and then apply 3D-DCT to integrate the energy in transformed domain. Watermarking signals are constructed by some bigger DCT coefficients and registered in the database of IPR to obtain copyright protection. Original 3D mesh data are not needed when detecting watermark. Experimental results show the watermark constructed by presented algorithm can resist vertex reordering, translating, rotating, uniformed scaling, and show robustness against adding noise, mesh simplifying, mesh filtering, and adding a secondary watermark.

关 键 词：3维水印 网格水印 零水印 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]