A Study on Object Modeling for X-ray CT based Engineering

在本研究项目中，研究了用于基于X射线CT扫描仪的先进数字工程系统的建模方法。我们特别感兴趣的是从CT数据生成三角网格或其他形式的曲面信息，这些信息可以用于数字工程系统。这种网格生成的一种标准方法是所谓的“等高线”方法。在这个为期两年的项目的第一年，我们开发了一种同时生成不同材料部分之间的边界曲面的方法。可应用于汽车发动机机体等由铁、铝等材料制成的机械零件上。这种方法基于三维图像分析进行材料分割，并基于我们基于匹配立方体理论的创新轮廓绘制方法。在第二年，我们专注于薄板结构的建模技术。重点研究了从CT数据中提取内侧表面的方法。中间曲面是…更有利于设计人员设计薄板结构。因此，从CT图像中提取它们的需求很大。我们的方法是结合了体积建模和网格建模的混合方法。我们首先提取与内侧表面相关的体素。然后，我们应用我们的等高线算法为中间曲面生成一个三角形网格。由于还可以提取薄板结构的边界面，因此可以计算薄板的厚度分布。在我们的工业应用中，准确性是相当重要的。我们设计和制造特定的样品(体模)来检查我们内侧表面的准确性。我们的实验表明，内侧曲面的精度与边界曲面的精度一样好。此外，我们正在研究焊接件的问题。薄板结构通常是由几块薄板焊接在一起构成的。使用X射线CT，不可能从图像中识别这样的焊接件。然而，在数字化工程应用中，焊接前每个零件的网格划分和有关焊接位置的信息是很重要的。因此，我们研究了一种将复杂焊接组件的CT数据分解为零件的方法。尽管还存在一些难题，但对于典型的焊接件，我们已经取得了成功。较少

In this research project, modeling methods to be used in advanced digital engineering systems based on X-ray CT scanner are investigated. Our particular interest is in generation of triangular meshes or other forms of surface information from CT data, which can be used in the digital engineering systems. One standard methods for this mesh generation is so-called "iso-contouring" method. In the first year of this 2 year project, we developed a method to generate boundary surfaces between different material parts simultaneously. It can be applied to such a mechanical part as automotive engine blocks which are made of iron and aluminum. This method is based on three dimensional image analysis for material segmentation and also on our innovative contouring method based on our "Matching Cubeablity" theory.In the second year, we were concentrated on modeling technologies for thin plate structures. Particularly we studied methods for extracting medial surface from CT data. A medial surface is … More useful for designers to design thin plate structures. So there is large demand to extract them from CT images. Our method is hybrid by combining volumetric modeling and mesh modeling. We first extract voxels which are related to the medial surface. Then we apply our contouring algorithm to generate a triangle mesh for the medial surface. Since we can also extract the boundary surface of the thin plate structure, thickness distribution of the thin plate can be evaluated. In our industrial applications, accuracy is rather important. We design and manufacture specific samples (phantom) to examine the accuracy of our medial surface. Our experiment shows that the accuracy of the medial surface is as good as that of the boundary surface.Furthermore we are working on the problem of welded parts. A thin plate structures are usually made of several pieces of thin plate which are welded together. Using X-ray CT, it is impossible to identify such welded part from the images. However, in digital engineering applications, meshes for each individual part before welding and information about the welding positions are important. So we studied a method to separate CT data of a complex welded assembly into its parts. Though there are still difficult problems, we have succeeded for typical welded parts. Less