基于曲率半径的数据分割

Segmentation of Measured Point Data Using Local Curvature Radius

数据分割是反向工程几何建模的关键一步，该文在对散乱点建立正交模型的基础上，利用参数曲面拟合法求解数据点的局部曲率，进而求出各点的局部平均曲率半径，根据曲率半径的阀值可以检测出各类边点，包括楞边、光滑过渡的峰线（ridge lines）和谷线（valley lines），然后进行边点连接形成封闭的边环，最后利用扫描线法把整体数据分为多个相互独立的子集。

： Segmentation is a key process in Reverse－engineering geometry modeling．This paper presents a segmentation method based on Orthogonal Cross－section Model（OCM）．At first，a parametric surface fitting method used to calculate local curvature is described．Second，various edge points including crease edge，ridge lines and valley lines are extracted．Third，edges are formed from edge points and edge loop produced from edges．At last，data is segmented by scan－line method．Real example is given．