摘要
为了有效解决高精度电子元器件表面缺陷提取精度偏低的问题,提出一种基于图像联合双边滤波的电子元器件表面缺陷视觉提取方法。该方法引入了机器视觉采集样本图像以及判断元器件表面图像有无缺陷,并构建SVM分类模型分类缺陷,将完成分类的电子器件表面光学图像作为输入,采用高斯函数计算获取图像的空间距离权值和灰度权值,将得到的权值相乘得到联合滤波值,构建联合双边滤波器。利用滤波器完成电子元器件表面缺陷图像滤波处理,并通过Gabor提取电子元器件表面缺陷。分析结果表明,所设计的方法可以有效提升缺陷提取速度和精度,并且平均交并比取值在0.96以上,具有良好的电子元器件表面缺陷提取性能。
In order to effectively solve the problem of low accuracy of surface defect extraction of high-precision electronic components,a visual extraction method of surface defect of electronic components based on image joint bilateral filtering is proposed.In this method,machine vision is introduced to collect sample images and determine whether the surface images of components have defects,and SVM classification model is constructed to classify defects.The optical images of the surface of electronic devices that have completed the clas-sification are taken as input,and the spatial distance weights and gray weights of the images are calculated by using Gaussian function.The obtained weights are multiplied to obtain joint filter values,and joint bilateral filters are constructed.The surface defects of electronic com-ponents are filtered by using filter,and the surface defects of electronic components are extracted by using Gabor.The analysis results show that the design method can effectively improve the speed and accuracy of defect extraction,and the average crossover ratio is above 0.96,which has good performance of surface defect extraction.
作者
王理想
邓宗强
WANG Lixiang;DENG Zongqiang(School of Artificial Intelligence,Guangdong Industry Polytechnic University,Guangzhou Guangdong 510399,China;Unmanned System Research Center,China People's Police University(Guangzhou),Guangzhou Guangdong 510663,China)
出处
《电子器件》
2025年第4期847-852,共6页
Chinese Journal of Electron Devices
基金
广东省教育厅2021年度普通高校重点科研项目(2021ZDZX1130)。
关键词
图像联合双边滤波
电子元器件
表面缺陷提取
图像处理
SVM分类器
提取精度
image joint bilateral filtering
electronic components
surface defect extraction
image processing
SVM classifier
extraction precision
