针对点云识别制造特征可以为逆向工程系统和CAX系统集成提供一个直接通道。利用特征组成面集及其之间的约束关系进行制造特征识别。首先利用模糊c-均值聚类方法对点云进行分区,利用分区过程中得到的隶属度找出面片的邻接关系,对分片点云进行曲面类型判别,利用最小二乘法进行平面或二次曲面的几何参数提取,对选取的面片进行平行、垂直、相切、凸凹等约束关系判别,与预先定义的特征规则相匹配,从而实现制造特征识别。实例验证了本文算法的有效性。
Manufacturing features recognition based point cloud data can offer a direct path for integration of reverse engineering and CAX system. Manufacturing features are recognized by a number of faces and relationships between these faces. Segmentation is realized by fuzzy c-means clustering method and adjacency relationships are found by memberships that are acquired in the process of segmentation ; surface types are distinguished and geometric parameters of planes or quadric surfaces are extracted by least square algorithm. Constraint relationships such as parallel, perpendicular, tangent, convex or concave are identified among selected patches. Features are recognized by matching these relationships with feature rules that are defined in advance. An example is presented to validate this approach.