中文摘要：

为满足有限元分析的需要，针对参数曲面提出一种几何自适应的网格生成方法．通过黎曼度量控制下的曲面约束Delaunay三角化获得曲面中轴，将其用于自动识别曲面邻近特征，并通过曲率计算自动识别曲率特征；根据邻近特征和曲率特征，融合传统网格尺寸控制技术控制边界曲线离散，并创建密度场；结合映射法和前沿推进技术对组合参数曲面生成几何自适应的网格．实验结果表明，该方法能够处理复杂的几何外形，生成的网格具有很好的自适应效果和质量．

英文摘要：

To satisfy the requirement of finite element analysis, a geometry adaptive mesh generation approach is proposed for parametric surface. Based on constrained Delaunay triangulation under Riemannian metric context, the new approach could capture the proximity between vertices and curves from surface boundary. Together with curvature and other traditional size control strategies, the proximity is used to control the discretization of surface boundary and construct the mesh size field. An advancing front technique based on mapping transformation is then adopted to discretize the surface. Experiments on complex industrial models show that this approach is efficient to generate high-quality adaptive surface mesh based on geometric features.