In this study, an oblivious 3D mesh watermarking scheme is represented utilising local curvature estimation and statistical characteristics of 3D mesh to provide robustness as well as retaining the imperceptibility of the 3D model. The proposed method estimates the local curvature of 3D model by finding the difference between the average normal and the surface normal of all the faces in a 1-ring neighbourhood of a vertex under consideration. Feature vector of all vertices is then measured and used to select vertices for watermark insertion. Distributions of vertex norms are transformed statistically to hide the watermark as statistical parameters are more robust and less prone to attacks. The robustness and imperceptibility of the proposed method against various attacks are analysed through simulations.

中文翻译：

---

使用局部曲率估计的3D网格统计水印方法

在这项研究中，通过利用局部曲率估计和3D网格的统计特性来表示一种遗忘的3D网格水印方案，以提供鲁棒性并保留3D模型的不可感知性。所提出的方法通过找到所考虑的顶点的1环邻域中所有面的平均法线和表面法线之间的差异来估计3D模型的局部曲率。然后测量所有顶点的特征向量，并将其用于选择水印插入的顶点。统计参数对顶点范数的分布进行统计转换，以隐藏水印，因为统计参数更健壮且更不容易受到攻击。通过仿真分析了该方法对各种攻击的鲁棒性和不易察觉性。