基于逆向工程的车身测量点云分割方法研究

主要研究实现了两种车身测量点云数据逆向建模过程中点云分割的算法——基于平面度的直接分割方法和基于对数法矢值的曲面分割方法。前者通过最小二乘拟合平面法矢量夹角的均方差值来搜索具有几何相似特性的连续曲面片。后者以起始点法矢量与其二阶邻域点的法矢量差的模为特征求解一对数值作为数据点的分割特征。最后结合车身曲面点云数据,给出不同的实例说明两种方法在车身逆向建模中针对不同特性曲面的分割有效性。

Segmentation for Point Cloud of Auto body Surface Reconstruction

This paper researches and realizes two algorithms of point clouds segmentation—the direct segmentation method based on planarity and the surface segmentation method based on logarithmic normal vector.The former searches continuous surface patches with geometric similarity,by using the mean-square deviation of vector contained angle of the least-square fitting planar method.The latter solves a pair of numerical number as the segmentation characteristic of data points,charactered by the modulus of normal vector of the initial points and normal vector difference of the second-order neighboring points.With the point clouds from body surface obtained by ATOS measurement equipment,the paper presents different examples,and proves the availbality of two methods at different conditions in the body reverse modeling.