基于三维点云的机械加工精度自动检测方法
2019-04-23
作者：杨坤，李子宽，李嘉
单位：河海大学
摘要：本文提出了一种基于三维点云的零件精度自动检测方法，对传统ICP（迭代最近邻点）方法进行改进，将基于配准误差的点对权值添加到坐标转换参数和总体残差的迭代计算过程中，克服了误差点对配准的影响；根据误差分布假设检验原理设计加工精度快速评价方法，帮助质检人员快速识别问题零件，提高检测效率和精度；通过试验验证了改进ICP算法对局部误差点云配准的鲁棒性以及加工精度评价方法的准确性。
关键词：机械构件；三维点云；鲁棒配准；正态分布假设检验；ICP
中图分类号：TG80；TH162文献标志码：ADOI：10.3969/j.issn.1000-7008.2019.04.028
Automatic Detection Method of Machining Accuracy
Based on 3D Point Cloud
Yang Kun，Li Zikuan，Li Jia
Abstract:An automatic detection method for parts accuracy based on 3D point cloud is ploposed in this paper.The traditional ICP(Iterative Closest Point)method is improved.The pointtoweight based on the registration error is added to the iterative calculation process of the coordinate transformation parameters and the overall residual，which overcomes the influence of the error point on the registration.According to the error distribution hypothesis test principle，a rapid evaluation method of machining accuracy is designed，which can help quality inspectors to quickly identify problem parts and improve detection efficiency and accuracy.The robustness improved ICP algorithm for local error point cloud registration and the machining accuracy evaluation method are verified by experiments.
Keywords:mechanical component；threedimensional point cloud；robust registration；normal distribution hypothesis test；ICP