【作者】 陈杰；

【导师】 高诚辉； 何炳蔚；

【作者基本信息】 福州大学 ， 机械制造及自动化， 2011， 博士

【摘要】 三角网格模型具有简单、直观的特性使之成为当前模型显示中最常用的格式。特征建模因其能同时面向设计和制造过程，被广泛应用于模型重建过程。特征对产品的几何外形具有极为重要的控制作用，因而特征处理技术成为当前反求工程领域的关键技术，而最核心的部分是特征提取，它是后续处理技术的基础和前提。由于过渡特征特别是混合过渡特征无法采用单一的法矢或曲率信息进行有效提取，因此过渡特征提取技术研究是目前研究的难点和热点。根据三角网格模型混合过渡特征的不规则性和拓扑结构原理，提出了一种多法矢夹角阈值特征提取算法进行混合过渡特征提取研究。首先用一组能对模型进行区域分割的法矢夹角阈值进行特征提取，通过区域分割确定每个非特征区域的点集；对非特征区域进行最小二乘拟合精度比较，确定各个区域法矢夹角阈值；用各区域的最佳法矢夹角阈值确定整个模型的特征，实现混合过渡特征的准确提取并对三角网格模型进行区域分割。研究了一种空间生长填充三角网格孔洞修补改进算法，进行混合过渡特征区域和跨区域的孔洞修补。首先对网格模型孔洞边界进行自动识别，用空间网格生长法对孔洞区域进行网格填充；选择孔洞边界相关顶点作为采样点对新增孔洞顶点进行最小二乘位置调整；引入区域分割和特征边界曲线拟合方法进行混合过渡特征及跨区域的孔洞修补，为提高多区域孔洞的修补精度提供一种可行方案。针对模型简化过程中特征保护问题，采取基于区域分割的模型简化方法。根据简化需要，结合区域面片数量比例和区域曲率值，合理调整分配各个区域的简化量，减少曲率较大的特征区域简化量；采用加入面积权值的QEM模型简化算法对各个区域进行独立简化，实现模型简化的同时有效地保护特征。在以上研究的基础上，构建了“基于混合过渡特征的三维模型重建系统原型”，该系统可将各研究算法转化为三维模型相关可视化处理过程中。更多还原

【Abstract】 Triangular mesh model possesses the simple and intuitive characteristics,whichmakes it the most popular format of model shows.Feature reconstruction is widelyused in reverse engineering model reconstruction process,for it can faces bothdesign and manufacturing process. Feature has very important control effect ofgeometrical shape of products, thus feature treatment technologies are the keyfield technology in reverse engineering, and the core part is feature extractiontechnique because this process is the foundation and prerequisite of follow-upprocessing. As the blend features especially mixed blend feature can not beextracted effectively by using single vector or curvature information, therefore, theblend feature extraction technology is a hotspot and quite difficult in currentresearch.According to irregularities and topological structure of mixed blend feature intriangular mesh model, this paper presents a new multi threshold of vector anglefeature extraction algorithm to hybrid blend feature extraction. Firstly，mixedblend feature is extracted by a group threshold angles of normal vector which cansegment the model preliminary, and point set with different threshold angles ofeach non-feature region are determined; then each region least-square fittingprecision of different threshold angles are compared in order to choose the bestthreshold angle of normal vector;Finally, mixed blend features are extractedaccurately by the best threshold angle of different regions, in this way trianglemesh model segmentation based on features is taken out.An improved space mesh growing hole filling algorithm is proposed aiming attriangular mesh holes in mixed blend feature region and cross regions. Firstly, allboundary feature is recognized and hole boundaries are automaticly identified byhole boundary properties; then the hole area is filled by meshes through spatialmesh growth method; finally, new added vertex positions are adjusted byleast-square method according to the boundary area vertex. Introducing region segmentation and boundary features curve fitted into mixed blend feature andcross regions hole filling algorithm, provides a feasible scheme for improvingfilling precision in multi-regions hole filling problem.For the purpose of feature regions protection in triangular mesh simplificationprocess, a mesh simplification algorithm based on region segmentation is adopted.According to the needs of simplification, regions simplified quantity distributionsare rationally adjusted by region mesh quantity ratios and region curvature values,and as a result, the simplified quantities of feature regions with large curvature arereduced. Each region is simplified independently by a QEM model simplificationalgorithm with added area weights of relevant triangle meshes; realize effectivelyfeature region protection in the triangle mesh model simplification processing."Processing prototype system based on mixed blend features of3D modelreconstruction" is constructed based on above studies.This system can transfereach algorithm into visualization in correlation model processing process.更多还原