基于多尺度分析的逆向工程截面线特征分割  

Feature segmentation of sectional curve based on multi-scale analysis in reverse engineering

作　　者：杨红娟[1] 周以齐[1] 陈成军[1] 

机构地区：[1]山东大学机械工程学院虚拟工程研究中心,山东济南250061

出　　处：《计算机集成制造系统》2007年第2期375-380,共6页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金资助项目(50175063);山东省自然科学基金资助项目(Y2006F12)。~~

摘　　要：为解决基于截面线特征重构中,初等曲线特征元和次要曲线特征元的分割问题,提出一种将多尺度分析与特征识别相结合的逆向工程特征分割方法。采用曲率尺度空间,实现截面线的多尺度分析和特征检测;利用相对转角法,实现单尺度特征融合;基于多尺度间信息传递和融合的多尺度特征融合算法获得多尺度特征集成。用多尺度曲率分析代替传统单一尺度逆向工程截面线分析,保证了特征分割结果,以及与基于特征设计和建模的特征元的一致性。合成复合曲线和实际扫描截面线的实例表明,提出的方法可以有效地实现逆向工程截面线多尺度特征分割。To obtain segmentation of primary curve primitives and secondary curve primitives for feature--based reconfiguration of sectional curves, a feature segmentation method was presented based on theory of multi-- scale and characteristics identification for reverse engineering. Curvature scale space was used to conduct multi--scale analysis and feature detection in the proposed method. Relative angle was used to realize single--scale feature fusion. Based on information transmission and fusion among multi--scale, multi--scale feature fusion algorithm was proposed to merge the multi--scale features. Comparing to conventional single--scale analysis of sectional curve in reverse engineering, multi--scale curvature analysis guaranteed the consistency between feature segmentation results and primitives for feature--based design and modeling. Synthetic curves and practical scanned sectional curves indicated that the proposed method was effective in multi--scale feature segmentation.

关 键 词：逆向工程 曲率尺度空间 特征识别 特征融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]