【作者】 段亮亮；

【导师】 贾超；

【作者基本信息】 燕山大学 ， 计算机应用技术， 2011， 硕士

【摘要】 借助计算机断层扫描技术可以很容易得到一系列含有内部缺陷信息的二维图像,在某些应用方面具有很大的局限性,因为它需要借助相关工程人员的空间想象以及经验来判断内部缺陷的三维形态。在实际应用中为进一步精确确定内部缺陷的密度分布、形状、空间位置以及尺寸等信息,需要将这些断层图像重建为三维模型。本文以此为背景,对重要构件内部裂缝缺陷断层图像的三维重建算法进行比较细致的研究。首先,针对传统表面重建未考虑相邻切片图像信息的关联性,提出了特征跟踪的表面重建算法。区域预测、区域变形和表面绘制是实现裂缝缺陷快速重建的关键。预测是根据当前切片图像区域对下一个切片图像的区域进行预估。由于预测的结果和实际区域不一致采用区域变形技术得到实际的区域。最后对得到的区域进行表面绘制完成三维重建。本算法在很大程度上减少不必要的检测,为大规模数据三维重建提供可能,尤其是裂缝区域只占整个体数据中很少一部分。其次,针对目前大部分表面重建方法无法得到缺陷内部信息的缺点,提出了一种新的基于有限元网格的方法,首先通过图像分割实现裂缝区域体素的提取,然后对生成的体素进行网格的构建,最后通过平滑处理消除网格之间存在的阶梯状界面,实现了缺陷表面及内部的重建。最后,借助OpenGL函数对重建的物体进行三维显示,并构建出重建系统的整体框架,定义若干个核心的类。该重建系统提供了OpenGL支持的三维显示界面、二维图像预处理结果的显示界面以及所有断层图像预览的显示界面等。更多还原

【Abstract】 A series of 2D images which comprise the internal defect information can be obtained with CT scan easily, In some applications, this technology has significant limitations, because it needs the space-related imagination and experience of engineer to determine the three dimensional shape of internal defects. In practice, further precise determination of the density distribution within the defect, shape, spatial position, size and other information, you will need three-dimensional model of these slices images. This paper takes this as a background and carries on a series of research for the 3D reconstruction algorithm of slice-images for internal crack defect of important component systematically.Firstly, due to traditional surface reconstruction does not consider the relevance of adjacent slice images, we propose the surface reconstruction algorithm using feature tracking. Regional prediction, regional deformation and surface rendering are the key to the reconstruction of crak defect. Prediction is intended to estimate the region of following slice images based on the current image. Due to the difference between the predicted region and the actual region we can get the actual region through region morphing. Finally, the region is applied surface rendering to complete the Three-dimensional reconstruction. The algorithm can reduce unnecessary testing, and make the three-dimensional reconstruction of large-scale data possible especially when the crack region is a fraction of the entire body of data.Secondly, as most of current surface reconstruction methods can not obtain the inner information of defect, we propose a new method based on finite element mesh. The first step, extract the voxels of crack regions through image segmentation. Secondly, generate the mesh using the voxels extracted from the last step. Finally eliminate the jagged interfaces through surface smoothing. Thus, we can reconstruct the whole cracks, including surface and internal.Finally, reconstruction object is displayed by OpenGL. This paper constructs the whole framework of system, and defines some pivotal classes. The reconstruction System provides several browser windows, including, OpenGL support for the 3D interface, 2D images preprocessing interface, displaying all slice-images interface.更多还原