摘要
针对冲压件表面缺陷人工检测效率低、误检率高以及劳动强度大等问题,提出了一种基于双打光模板匹配的冲压件表面缺陷检测方法。首先针对不同的表面缺陷类型,根据照明效果采用底部环形光源与顶部面光源两种不同的打光方式。接着先后采用边缘提取、旋转定位、填充、模板匹配及缺陷提取等方法对采集的冲压件图像进行处理。在线检测应用表明,划痕、月牙磕碰缺陷采用低角度环形光源的打光方式和锈迹缺陷采用高角度面光源的打光方式具有良好的缺陷提取效果。整个表面缺陷图像处理方法可以快速、有效地检测出冲压件的表面划痕、月牙磕碰、锈迹等缺陷。检测精度最高可达0. 134 mm2,检测速度最快可达1. 196 s。该方法在检测速度和精度上能满足工厂检测要求,大幅提高生产效率,降低生产成本。
For the problems of low manual inspection,high false detection rate and labor intensity in the surface defects of stamping parts,based on double-shot template matching,a surface defect detection method of stamped parts was proposed. According to the different surface defect types,two different lighting patterns,such as a bottom ring light source and a top surface light source,were adopted based on the lighting effect,and the collected stamping parts images were processed by edge extraction,rotation positioning,filling,template matching,and defect extraction. The application of online inspection shows that for the scratching and crescent bump defects,good defect extraction effect is obtained by low-angle ring light source lighting,and for the rust defects,high-angle surface light source lighting method is applied. Therefore,the entire surface defect image processing method can quickly and effectively detect defects such as surface scratches,crescent moon bumps,and rust marks on the stamped parts. Detection accuracy is up to 0. 134 mm2,and detection speed is up to1. 196 s. Furthermore,this method can meet the requirements of factory inspection in terms of detection speed and accuracy to greatly increase production efficiency and reduce production costs.
作者
李松
周亚同
张忠伟
池越
韩春颖
Li Song;Zhou Yatong;Zhang Zhongwei;Chi Yue;Han Chunying(School of Electronics and Information Engineering,Hebei University of Technology,Tianjin 300401,China;Beijing Anshizhongdian Technology Co.,Ltd.,Beijing 100871,China)
出处
《锻压技术》
CAS
CSCD
北大核心
2018年第11期137-145,共9页
Forging & Stamping Technology
基金
河北省引进留学人员资助项目(CL201707)
教育部人文社会科学研究规划基金(15YJA630108)
关键词
冲压件
表面缺陷
检测
打光方式
模板匹配
检测精度
检测速度
stamping part
surface defects
detection
lighting method
template matching
detection accuracy
detection speed