Automatic Data Segmentation and Geometric Reasoning of Unorganized Point Cloud for Reverse Engineering of Precision Mechanical Objects

用于精密机械物体逆向工程的无组织点云的自动数据分割和几何推理

• 批准号: 0100074
• 负责人: Chia-Hsiang Menq
• 金额: $ 28.93万
• 依托单位: Ohio State University Research Foundation -DO NOT USE
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2001
• 资助国家: 美国
• 起止时间: 2001-07-01 至 2006-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0100074&HistoricalAwards=false
• 关键词: Automatic; Data; Segmentation; Geometric; Reasoning

This grant provides funding for the development of the necessary technologies that support automatic geometric model reconstruction of existing objects having arbitrary topology when the prototype is created or modified on the shop floor and when a computer-aided design model does not exist. Particularly, the focus is to automate the data segmentation process on the reconstructed manifold surface and to facilitate optimal geometric model reconstruction for precision mechanical parts by geometric computation and reasoning. An algorithm will be developed to efficiently reconstruct a triangle mesh from 3-dimensional unorganized coordinate points to optimally recover the object shape. The reconstructed triangle mesh captures object topology with the associated 2-manifold represented as a combinatorial structure, which establishes explicit relations among the data points, and provides a topological domain with necessary differential geometric characteristics for the subsequent data segmentation process. By extending segmentation concepts from the regular image domain to the irregular mesh domain, a robust two-step automatic data segmentation approach will be developed, combining the border-based approach and the region growing approach. Finally, algorithms for geometric computation and reasoning will be developed to automatically classify the segmented patches into surface elements and to infer possible topological relations and geometric constraints among them. If successful, the results of this research will lead to significant improvements in automatic geometric model reconstruction and reverse engineering. With automatic data segmentation and intelligent geometric reasoning, user intervention can be eliminated and the entire model reconstruction process can be shortened from days to minutes. When integrated with state of the art scanning devices, the developed technologies could lead to a seamless reverse engineering process and support rapid design and prototyping of high-precision mechanical components. The results will have potential application in a whole spectrum of engineering problems with a major impact on rapid design and prototyping, shape analysis, and virtual reality.

该赠款为开发必要的技术提供资金，这些技术支持在车间创建或修改原型以及不存在计算机辅助设计模型的情况下自动重建具有任意拓扑结构的现有物体的几何模型。 特别是，重点是自动化的数据分割过程中重建的流形表面，并促进最佳的几何模型重建精密机械零件的几何计算和推理。 将开发一种算法，以有效地重建一个三角形网格从三维无组织的坐标点，以最佳地恢复对象的形状。 重建的三角形网格捕捉对象的拓扑与相关的2流形表示为一个组合结构，建立显式的数据点之间的关系，并提供了一个拓扑域的必要的微分几何特征，为后续的数据分割过程。 通过将分割概念从规则图像域扩展到不规则网格域，结合基于边界的方法和区域生长方法，将开发一种鲁棒的两步自动数据分割方法。 最后，几何计算和推理的算法将开发自动分类的分割补丁到表面元素，并推断可能的拓扑关系和几何约束。 如果成功，本研究的结果将导致自动几何模型重建和逆向工程的显着改善。 通过自动数据分割和智能几何推理，可以消除用户干预，整个模型重建过程可以从几天缩短到几分钟。 当与最先进的扫描设备集成时，所开发的技术可以实现无缝的逆向工程过程，并支持高精度机械部件的快速设计和原型制作。 结果将有潜在的应用在整个频谱的工程问题的快速设计和原型，形状分析和虚拟现实的重大影响。