三角网格模型的特征线提取

在反求工程中，散乱数据点云的曲面重构常采用三角网格模型，若将其转换成曲面实体模型则有更广泛的应用，从三角网格模型中提取特征线是转换过程中的重要步骤．在讨论反求工程中数据点云分块方法的基础上，采取“基于边”的方法来提取特征线：先提取特征点，再连接成特征线．根据相邻三角片的法矢夹角和各点主曲率是否为极值，分两次提取特征点，利用三角顶点加权和均匀化等方法减少狭长三角片对特征点提取的计算误差影响，再将特征点分组连接成B样条曲线．文中算法的结果可为B样条曲面分片拟合和建立B-rep曲面实体模型提供依据。

Feature Line Extraction from Triangular Mesh Model

Feature lines such as boundaries and ridges should be preserved in transforming triangular meshes into rectangular B-spline patches. An edge-based algorithm that includes two steps: feature point detection and feature line generation is presented. The feature points are identified by the angle between the normals of adjacent triangles as well as by the principal curvatures. In the algorithm, weights and some other measures are applied to reduce the approximation error. Feature points are linked into B-spline curve for the subsequent construction of B-rep model.