计算机集成制造系统 ›› 2020, Vol. 26 ›› Issue (第2): 356-365.DOI: 10.13196/j.cims.2020.02.008

• 当期目次 • 上一篇    下一篇

激光刻写零件瑕疵图像在线检测算法

雷泰,柳宁+,李德平,王高   

  1. 暨南大学信息科学技术学院
  • 出版日期:2020-02-29 发布日期:2020-02-29
  • 基金资助:
    国家自然科学基金资助项目(61775172);广东省省级科技计划资助项目(2017B090910012) 。

Image online detecting algorithm for defects of laser writing parts

  • Online:2020-02-29 Published:2020-02-29
  • Supported by:
    Project supported by the National Natural Science Foundation,China(No.61775172),and the Guangdong Provincial Science and Technology Plan,China(No.2017B090910012).

摘要: 针对复杂背景条件下激光刻写零件瑕疵的机器视觉检测需求,提出一种基于导向线的刀路轮廓质量检测方法。先根据矢量文件生成导向线,对导向线中的线路与面域进行区分,再与待测图像配准,在导向线上生成采样点,最后对已滤波图像上轮廓的线宽、连通性和毛刺情况进行检测。有别于传统灰度或轮廓的差分检测方法,所提方法以导向线为路径,利用检测模板对采样点所在的局部图像进行检测。实验结果表明,该方法在复杂背景条件下对轮廓检测既具有良好的鲁棒性,又能够保证高检测精度与较好的实时性,满足激光刻写零件过程中的质量检测要求。

关键词: 机器视觉, 缺陷检测, 矢量导向线, 线性轮廓, 激光刻写

Abstract: To meet the defects detection demand of laser writing parts for machine vision under complex background conditions,a method of tool path contour quality detection based on guide lines was proposed.The guide lines were generated from the vector file,which were distinguished from the area.The matching operation was performed on the measurement-needed image with the generated guide lines,and  the sampling points was generated along the guide lines.A detection operation was performed on the width of contour,connectivity and burr information of the filtered image.Different from the traditional difference detection method performed on gray value or contour,the suggested method treated the guide line as the path,and performed detection operation on the local image of sampling points using detection template.The experimental results revealed that the suggested method both had a good robustness under the complex background condition,and could ensure high detection accuracy and good real-time performance and meet the quality inspecting requirements in the process of laser parts writing.

Key words: machine vision, defect detection, vector guide line, linear contour, laser writing

中图分类号: