锥台工件点云拟合算法研究  

Research on Point Cloud Fitting Method for Cone Workpieces

作　　者：王佳裕 李培波[2] 赵言正[1] WANG Jiayu;LI Peibo;ZHAO Yanzheng(Research Institute of Robotics,Shanghai Jiao Tong University,Shanghai 200240,China;College of Mechanical Engineering,Donghua University,Shanghai 201620,China)

机构地区：[1]上海交通大学机器人研究所,上海200240 [2]东华大学机械工程学院,上海201620

出　　处：《机床与液压》2023年第10期1-6,共6页Machine Tool & Hydraulics

摘　　要：针对不完整锥面工件模型拟合问题背景,为了解决大量点云数据的标准几何模型拟合问题,提出一种基于遗传算法的锥面点云数据拟合算法:通过随机选取拟合模型采样点,用点云数据点-标准模型采样点的距离计算替代点云数据点-标准曲面模型的距离运算,提高计算效率同时保留点云细节特征;通过对点云数据建立Kd-tree索引提升标准模型采样点距离检测效率;通过遗传算法迭代搜索拟合模型参数,并通过变数据量、变拟合精度等方法改进遗传算法的搜索效率。经实验验证:采用该算法可有效提升部分锥面工件拟合的匹配精度,误差范围±0.5 mm,且对于大量点云数据的处理效果良好。Aiming at the background of incomplete cone workpiece model fitting problem,in order to solve the standard geometric model fitting problem of a large number of point cloud data,a fitting algorithm for cone point cloud data based on genetic algorithm was proposed:by randomly selecting the fitting model sampling points,the distance calculation of point cloud data points-standard model sampling points was used to replace the distance calculation of point cloud data points-standard surface model,so as to improve the calculation efficiency and retain the detailed characteristics of point cloud;by establishing Kd-tree index on point cloud data,the distance detection efficiency of sampling points of standard model was improved;the parameters of the fitting model were searched iteratively by genetic algorithm,and the search efficiency of genetic algorithm was improved by changing the amount of data and fitting accuracy.The experimental results show that the algorithm can effectively improve the matching accuracy of some conical workpiece fitting,with an error range of±0.5 mm,and has a good effect on the processing of a large number of point cloud data.

关 键 词：点云拟合 KD-TREE 遗传算法 反求工程 

分 类 号：TH164[机械工程—机械制造及自动化]