基于机器视觉多目标工件分类识别和定位研究  

Research on Classification,Recognition and Localization of Multi-objective Workpiece Based on Machine Vision

作　　者：杨桂华[1] 刘志毅 王晓文 YANG Guihua;LIU Zhiyi;WANG Xiaowen(College of Mechanical and Control Engineering,Guilin University of Technology,Guilin Guangxi 541004,China)

机构地区：[1]桂林理工大学机械与控制工程学院,广西桂林541004

出　　处：《机床与液压》2021年第11期82-86,126,共6页Machine Tool & Hydraulics

基　　金：国家自然科学基金地区科学基金项目(61863009)。

摘　　要：为实现对随机摆放的多类型工件进行分类识别和定位,提出一种基于连通域Blob分析与神经网络分类器相结合的方法。该方法运用机器视觉技术,以Halcon软件为试验检测平台,通过Blob分析提取工件的特征信息实现定位,并且应用提前训练好的神经网络分类器对多类型工件进行分类识别。试验结果表明:相比传统模板匹配和定位算法,该多工件分类识别和定位方案识别准确率达到100%,定位精度提升0.7 mm,识别和定位时间减少8.735 ms,具有更好的多工件识别定位效果,有一定的应用前景和推广价值。In order to recognize and locate the multi-type workpiece placed randomly,a method based on connected domain Blob analysis and neural network classifier was presented.Machine vision technology was used and Halcon software was taken as the experimental platform,the feature information of the workpiece was extracted through Blob analysis,the neural network classifier trained in advance was used to classify and recognize the multi-type workpiece.The experimental results show that compared with the traditional template matching and localization algorithm,the recognition accuracy of the multi-workpiece classification and localization scheme is 100%,the localization accuracy is improved by 0.7 mm,and the recognition and localization time is reduced by 8.735 ms,to achieve a better multi-workpiece identification positioning effect.It has a certain application prospect and promotion value.

关 键 词：机器视觉 神经网络分类器 多工件识别定位 

分 类 号：TP242.2􀆰[自动化与计算机技术—检测技术与自动化装置]