成品电解镍表面质量自动检测  

Automatic Detection of Surface Quality of Finished Electrolytic Nickel

作　　者：陆迎东 许平[1,2] 朱正云[1] 王磊 高双飞 LU Yingdong;XU Ping;ZHU Zhengyun;WANG Lei;GAO Shuangfei(Faculty of Mechanical and Electrical Engineering,Kunming University of Science and Technology,Kunming 650500,China;Jinchuan Nickel and Cobalt Research and Design Institute,Jinchang 737100,Gansu,China)

机构地区：[1]昆明理工大学机电工程学院,昆明650500 [2]金川集团股份有限公司,甘肃金昌737100

出　　处：《机械科学与技术》2025年第1期150-156,共7页Mechanical Science and Technology for Aerospace Engineering

摘　　要：针对成品电解镍表面的气孔和结粒依靠人工进行检测时存在检测精度差、效率低等问题,提出一种基于点云的成品电解镍表面质量自动检测算法。该算法采用一种基于体素下采样优化的RANSAC平面分割和孔洞识别算法,对于数据量较大的电解镍点云集,首先通过体素过滤器返回下采样的点云,在不影响结构特征的前提下降低了数据的冗余度,然后采用RANSAC算法进行平面拟合,平面拟合的外点即结粒点云,对得到的结粒点云采用欧式聚类法进行聚类,再对聚类后的点云根据结粒高度进行合格与不合格分类;最后根据平面拟合得到的平面系数将所有点进行投影,以每一个点为原点建立坐标系,找到每个点的k近邻点,计算k近邻点在该点坐标系下的有向角,判断其是否满足预先设定的条件,判定目标点是否为孔洞点。实验结果表明:该方法在有效减少计算量的同时,能够获得较高的检测精度。Aiming at the poor accuracy and low efficiency of manual identification of electrolytic nickel,a point cloud based defect recognition algorithm for electrolytic nickel surface is proposed.The algorithm adopts a plane segmentation and hole identification algorithm based on the voxel down sampling optimization.For electrolytic nickel point clouds with large amount of data,the down sampled point cloud is firstly returned through voxel filter,which reduces the redundancy of data without affecting the structural characteristics.Then,the algorithm is used for plane fitting.The outer points of plane fitting are granulation point clouds.The obtained granulation point clouds are clustered by using the European clustering method.Then the clustering point cloud is classified as qualified and unqualified according to the grain height.After that,all points are projected according to the plane coefficient obtained by using the plane fitting,the coordinate system is established with each point as the origin,and then the k-nearest neighbour point of each point is found.The directional angle of the k-nearest neighbour point in the point coordinate system is calculated to determine whether it meets the preset conditions and whether the target point is a hole point.The experimental results show that the present method can effectively reduce the amount of calculation and achieve high recognition accuracy.

关 键 词：点云 表面质量 平面拟合 欧式聚类 有向角 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]