三维CAD模型相似性比较中姿态调整算法

3D CAD Model Pose Adjustment Algorithm for Similarity Assessment

三维CAD模型相似性评价在工程中具有重要的意义,为达到该目的,常常需要先对模型进行姿态调整。三维模型的姿态调整是指将两个坐标系下建立的模型调整到统一的坐标系下,以便于实现模型的相似性比较。主成分分析法(Principal Component Analysis,PCA)是最常用的模型姿态调整算法。但传统的方法没有考虑坐标轴方向不同的选取对算法效果的影响。描述了运用改进的PCA算法对相比较的两个CAD模型进行姿态调整的详细步骤和方法,并讨论了PCA算法产生的坐标系中各坐标轴方向的确定方法。最后,运用算例证明了文中所给出的算法和结论的正确性。

3D CAD model similarity has important applications in engineering. In order to achieve this purpose, it is needed to adjust the pose of the models. 3D model pose estimation is often performed by Principal Component Analysis （PCA）. This transformation changes the coordinate system axes to new ones which coincide with the directions of the three largest spreads of the point distribution. The algorithm, however, has not considered that different selection for the axes directions has impacted on the effectiveness. So, it will appear to be unstable for some models. It describes the detailed steps and methods using the improved PCA algorithm to achieve pose registration for the two models. The axis direction of the coordinate systems which is obtained by the PCA approach is discussed. Finally, experimentation shows the correctness of the method and conclusion.