首页 | 本学科首页   官方微博 | 高级检索  
     

基于区域分割与约束逼近的二次曲面重建
引用本文:柯映林,单东日.基于区域分割与约束逼近的二次曲面重建[J].浙江大学学报(自然科学版 ),2004,38(10):1274-1279.
作者姓名:柯映林  单东日
作者单位:[1]浙江大学机械与能源工程学院,浙江杭州310027 [2]山东轻工业学院机械与电气学院,山东济南250100
基金项目:教育部高校骨干教师资助计划,高等学校博士学科点专项科研项目
摘    要:从散乱数据中有效地分离出特征与约束是表达原始产品设计意图的关键.基于点的连通性及同一特征曲面数据点估算法矢、曲率的相似性,并在特征曲面拟合误差控制下实现了空间散乱数据二次曲面数据的自动分离;应用二次曲面的几何参数表达建立了二次曲面整体约束逼近的最优化数学模型,将几何约束情况下的二次曲面重建转化为数学模型的求解,使重建的CAD模型可有效再现二次曲面之间的几何约束.工程应用实例表明:该方法稳定可靠,可显著提高反求工程CAD建模效率及重建模型精度.

关 键 词:反求工程  约束优化  二次曲面  区域分割
文章编号:1008-973X(2004)10-1274-06
修稿时间:2003年10月30

Quadric surface reconstruction based on segmentation and constrained optimization
KE Ying-lin,SHAN Dong-ri.Quadric surface reconstruction based on segmentation and constrained optimization[J].Journal of Zhejiang University(Engineering Science),2004,38(10):1274-1279.
Authors:KE Ying-lin  SHAN Dong-ri
Abstract:Quadric surface was one of the most accepted methods for representing the complex shapes at (industrial) products,due to its natural properties.In raverse engineering, how to extract the valid features and its intrinsic topological constraints from unorganized data is the key to racover the original design intendt.A new algorithm for constrained (reconstruction) of quadric surface from unorganized 3D points was proposed. The method includes two steps. First, based on the adjacent relationship and similarity of normal vector or curvature of points, points distributed on quadric surface were segmented from the measured points-cloud with region growing method under a given error for surface fitting. Second, for multiple surfaces fitting subject to geometric constraints, a general mathematical model was established. By solving the model, the regularities and symmetries applied in product design originally can be represented and used to reconstruct the CAD model. Engineering applications show that the precision and efficiency of remodeling are (improved) remarkably.
Keywords:reverse engineering  constraint optimization  quadric surface  region segmentation
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号