轴承套圈端面缺陷在线视觉检测的研究与实现  

Research and Implementation of On-Line Visual Inspection of Defects on End Face of Bearing Rings

作　　者：陈硕[1] 林志敏 吴岳彬 钟原 应铭 CHEN Shuo;LIN Zhimin;WU Yuebin;ZHONG Yuan;YING Ming(School of Mechanical Engineering and Automation,Fuzhou University,Fuzhou 350116,China;Fujian Yongan Bearing Co.,Ltd.,Yongan 366000,China)

机构地区：[1]福州大学机械工程及自动化学院,福州350116 [2]福建省永安轴承有限公司,福建永安366000

出　　处：《轴承》2022年第2期48-54,共7页Bearing

摘　　要：针对轴承生产企业套圈生产过程中普遍存在端面缺陷的问题与人工目检的现状,提出了基于机器视觉的轴承套圈端面缺陷在线检测方法。首先,对套圈图像预处理后进行边缘检测,采用四连通域定位套圈端面区域;其次,采用最小二乘法拟合端面轮廓以判别外形缺陷,采用极坐标变换将套圈环形端面拉伸成矩形,采用Sauvola局部二值化算法对矩形图进行缺陷分割,并通过坐标系反变换与双线性插值法将缺陷图转换回环形图;最后,根据提取缺陷的图像特征完成缺陷的识别与分类。现场测试表明,套圈端面检测系统的整体识别准确率达98.6%。Aimed at common problems of defects on end face during production process of rings produced by bearing enterprises and status of manual visual inspection,an on-line inspection method for defects on end face of bearing rings based on machine vision is proposed.Firstly,the edge inspection is carried out after preprocessing of ring image,and the four adjacent connection domains are used to locate the ring end face area.Secondly,the least square method is used to fit the profile of end face to identify the shape defects.The ring end face is stretched into a rectangle by using polar coordinate transformation,the Sauvola local binarization algorithm is used to segment the defects of rectangular image,and the defect image is converted back into ring image through inverse coordinate system transformation and bilinear interpolation method.Finally,the identification and classification of defects are completed according to extracted image features of defects.The field test shows that the overall recognition accuracy rate of ring end face inspection system reaches 98.6%.

关 键 词：滚动轴承 套圈 表面缺陷 机器视觉 图像处理 

分 类 号：TH133.33[机械工程—机械制造及自动化] TG156[金属学及工艺—热处理]