基于ANN-NURBS的散乱数据点自由曲面重构

Approach of free-form surface reconstruction from scattered data based on artificial neural network and NURBS

探讨了曲面三维密集散乱点数据的几何建模方法。按照先压缩后拟合的两步方法重构策略,实施基于ANN-NURBS的散乱点自由曲面重构。提出了基于人工神经网络(ANN)的散乱数据点的拓扑矩形网格重建方法并建立了神经网络模型。该模型利用神经元对曲面散乱点的学习和训练来模拟曲面上的点与点之间的内在关系,结点连接权矢量集作为对散乱点集的工程近似化并重构曲面样本点的内在拓扑关系。算例表明,该方法可实现三维密集散乱点数据自组织压缩,生成期望疏密程度和精度的矩形拓扑网格,并可有效保持原数据点集的拓扑特征,从而实现了基于NURBS的大规模散乱数据点的精确曲面重构。

An approach is presented to model the 3D scattered data measured from the curved surfaces in two stages. The first stage employs the self\|organizing feature map neural network to extract the large scale scattered data and produce the topologic rectangular mesh. It follows the second stage by reconstructing the surface from the topologic rectangular mesh based on the NURBS method. The computer simulation shows that the large scale scattered data can be reduced to the reasonable scale, while the topologic features of the whole scattered data are kept, thereby the precise surface reconstruction with NURBS is realized.