一种基于NURBS的三维统计形状建模方法

A 3D statistical shape modeling method based on NURBS technique

针对当前主动形状模型构建中的实际问题,本文提出一种整合NURBS技术的三维统计形状建模方法。首先对目标样本进行多边形网格化处理,再对其进行NURBS建模和曲面空间变换,然后对NURBS曲面上的控制点进行主成分分析,以生成相应的统计形状模型。为了验证该方法的有效性,我们将该方法应用于部分人体器官的训练样本集进行建模测试。实验结果表明,该方法可较为合理和准确地描述目标形状的变化模式,并能提高建模效率,减少存储空间的占用。

According to the current existing problems of Active＇Shape Modeling（ ASM） ,a 3D statistical shape modeling method integrated with NURBS technique is presented in this paper. First, the target training sets was polygonized. Secondly, NURBS modeling and spatial transformation was implemented with the surface. After that, principal component analysis （PCA） was performed on the control paints of NURBS surface to build the statistical shape model. To evaluate the effectiveness of the method, it was utilized to apply for the statistical shape modeling of some human organs. The results shows that the method can describe the varation modes of the target reasonably and realistically, as well as improve the efficiency of modeling and reduce the memory storage.