基于RBF神经网络的三维网格实体模型重构的研究

A New Approach to Constructing 3-Dimensional Grid Solid Geometry Model Based on RBF Neural Networks

提出了一种新的基于RBF神经网络的重构三维网格实体模型的算法 ,输入是未知表面的采样点坐标集 ,输出是该未知表面的三维网格近似。同时提出了一个简便而有效的在三维域上三角形网格的Laplacian光顺造型方案 ,网格光顺的结果能保证重建的结果比较光滑。算法较传统的算法更精确和更可靠。

A novel scheme for constructing 3D grid solid ge om etry model is presented which is based on RBF(radial basis function )neural netw orks. The inputs are scattered 3D-data with specified topology. The outputs are 3D solid geometry model. A simple and novel smoothing scheme for triangular mes hes on a 3D non-planar surface region is also presented. The method can improve the robustness and reliability of the traditional approaches. The recovered 3D shape is then shown along with the original surface. In comparison with the trad itional methods, examples show that the algorithm is accurate and reliable.