Abstract: In this paper，a new digital watermarking scheme for a 3D shape model defined as a point set has been proposed.The algorithm is based on an incrementally expanding Neural Network and the statistical analysis of its learning process for an unorganized point cloud.Using the information obtained from the study of these normal variations of vertices near a sharp edge or a high curvature area of the target space to detect features.The watermarks embedded by using the method can be detected，to some extent，after such attacks as cropping，（simulated） simplification，similarity transformation，and additive random noise applied on the watermarked point set model.

Key words: digital watermarking, geometric modeling, neural network, feature detection

摘要： 对定义在一系列无方向点集上的3D形状模型提出一种新的数字水印方案。首先应用神经网络的方法对点集模型进行学习，根据在学习过程中神经网格法线的变化情况，找到特征点。最后对已加水印的点集模型应用不同的攻击，比如剪切、（模拟）简化和附加随在特征点附近嵌入水印信息机噪声之后，结果显示通过这种方案嵌入的水印，在某种程度上仍可以被检测到。

关键词: 数字水印, 几何建模, 神经网络, 特征点