摘要：

提出一种基于奇异值分解的网格模型盲水印算法，只对网格顶点的几何数据进行处理，适用于任意拓扑结构的网格模型。奇异值分解在与网格模型几何数据局部统计特征相似的球面坐标映射方阵中进行，水印序列嵌入到方阵生成的奇异值序列中。实验结果表明，算法可抵抗平移、旋转、各向一致缩放攻击及顶点重排序攻击，对噪声攻击也具有一定的鲁棒性。

关键词: 三维水印, 网格水印, 盲检测, 奇异值分解

Abstract: The blind watermark algorithm based on SVD proposed manipulates vertex’s geometric features, which can be used in meshes with any topological connectivity. SVD is processed in a 2D mapping grid, which is produced by vertex’s spherical coordinates and has similarly local statistical property of original mesh. Watermark sequence is embedded in singular value sequence. Experimental results show the algorithm can resist translation, rotation, uniformed scaling, vertex reordering, and is also robust against adding noise.

Key words: 3D watermark, mesh watermark, blind detection, singular value decomposition(SVD)

中图分类号: 

• TP309