基于几何矩不变量分析的视觉测力方法  

Visual force measurement method based on geometric moment invariants analysis

作　　者：胡文[1] 印波 张瑞 HU Wen;YIN Bo;ZHANG Rui(School of Digital Media,Shenzhen Institute of Science and Technology,Shenzhen 518172,China;School of Intelligent Manufacturing and Equipment,Shenzhen Institute of Science and Technology,Shenzhen 518172,China)

机构地区：[1]深圳信息职业技术学院数字媒体学院,广东深圳518172 [2]深圳信息职业技术学院智能制造与装备学院,广东深圳518172

出　　处：《传感器与微系统》2020年第10期27-30,共4页Transducer and Microsystem Technologies

基　　金：广东省自然科学基金资助项目(2018A030313382);深圳市科技计划资助项目(JCYJ20170306095702695,2017GKQNCX069)。

摘　　要：针对微装配过程中微力测量的问题,提出一种基于几何矩不变量分析的视觉测力方法。利用形状描述算子几何矩不变量构成特征向量,用于描述微夹爪变形前后的形状,推导出几何矩不变量特征向量与受力之间的映射关系,在此基础上建立测力模型及其求解方法;然后建立训练集,输入为在已知受力下微夹爪形状对应的特征向量,输出为已知受力;同样地建立测试集,输入为在未知受力下微夹爪形状对应的特征向量,输出为未知受力;最后支持向量机(SVM)比较测试集与训练集中的特征向量,对测试集的输入进行多类分类,从而测量未知受力。实验结果表明:该方法可以准确而稳定地测量微力。Aiming at the problem of force measurement in microassembly,a new visual force measurement method based on geometric moment invariants analysis is proposed.The method uses shape descriptor of geometric moment invariants to describe the microgripper shape before and after deformation.Mapping relationship between applied forces and feature vectors of geometric moment invariants is deduced.On this basis,set up force measurement model and its solution.Then,a training set is established,where the input represents the feature vector corresponding to shape of microgripper,and the output represents the corresponding known applied force,respectively.Similarly,a testing set is established,where the input represents the feature vector corresponding to shape of microgripper and the output represents the corresponding unknown applied force,respectively.Finally,a support vector machine(SVM)compares each feature vector in the testing set with those ones in the training set,implements multi-class classification,so as to measure unknown applied force.Experimental results show that the proposed method is accurate and stable.

关 键 词：微装配 视觉测力 几何矩不变量 支持向量机 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] TP391.4[自动化与计算机技术—控制科学与工程]