基于偏振视觉的轴承曲面划痕缺陷检测  

Bearing Surface Scratch Defect Detection Based on Polarization Vision

作　　者：韩鸿志 王福明[1] 刘育君 李江涛 赵鹏飞[1] HAN Hongzhi;WANG Fuming;LIU Yujun;LI Jiangtao;ZHAO Pengfei(School of Mechanical Engineering, North University of China, Taiyuan 030051, China;Taiyuan Unicom Branch, Taiyuan 030001, China;Information Processing Office of Comprehensive Support Center of Shanxi Armed Police Corps, Taiyuan 030012, China)

机构地区：[1]中北大学机械工程学院,山西太原030051 [2]太原联通分公司,山西太原030001 [3]武警山西总队参谋部综合保障中心情报处理室,山西太原030012

出　　处：《中北大学学报（自然科学版）》2022年第2期116-121,共6页Journal of North University of China(Natural Science Edition)

基　　金：国家自然科学基金资助项目(62175219)。

摘　　要：针对轴承工作曲面划痕缺陷成像的问题,提出了一种采用正交偏振成像的检测方法.首先,采用偏振相机结合正交偏振理论获取光滑轴承曲面图像,利用形态学滤波差分运算消除不均匀背景;然后,采用特定滤波和增强方式对SWT产生的分解系数进行保边去噪和增强,并进行图像重构;最后,对其中的缺陷特征进行分割处理.结果表明:采用线偏振光源照射光滑轴承曲面,局部过曝光现象消失,获得低对比度划痕缺陷特征.该方法可有效提高轴承工作曲面划痕缺陷的检测准确度,也可推广到类轴承光滑金属曲面划痕缺陷的视觉检测中.Aiming at the problem of scratch defect imaging of bearing working surface,a detection method based on orthogonal polarization imaging was proposed.Firstly,the image of smooth bearing surface was obtained by polarization camera combined with orthogonal polarization theory,and the uneven background was eliminated by morphological filtering and differential operation;Secondly,specific filtering and enhancement methods can denoise with edge-preserving smoothing and enhance the decomposition coefficients generated by SWT,and the image was reconstructed;Finally,the defect features were segmented.The results show that when the linear polarization light source is used to irradiate the smooth bearing surface,the local overexposure phenomenon disappears and the scratch defect characteristics of low contrast are obtained.This method can effectively improve the detection accuracy of bearing working surface scratch defects,and can also be extended to the visual detection of bearing like smooth metal surface scratch defects.

关 键 词：偏振成像 退偏振 形态学滤波 轴承 划痕检测 

分 类 号：TH744[机械工程—光学工程]