摘要
冰箱上的启动器在生产中常出现字符不合格、表面污渍、外壳破损、顶盖缺失、铁片弯折以及铁片缺失黄铜等缺陷。针对以上典型缺陷,首先结合生产实际设计并搭建外观缺陷检测系统,然后通过模板匹配以及灰度差分处理,对各个待检测区域进行分割,最后基于颜色特征、灰度均值、区域定位以及面积等特征参数建立模型对零件进行算法验证。试验结果表明,此算法能够对启动器复杂多样的缺陷进行检测,且成本较低、检测率较高以及降低误检率,对工业零件的外观检测有很好的实用推广价值。
In production process,the refrigerator starters often appear character failures,surface stains,shell damages,missing roofs,bending iron sheets missing brass,etc. In order to solve the above typical defects,the paper constructs the appearance defect detection system based on the production design. Then,through the template matching and the gray level difference processing,each area to be detected is divided. Based on the color feature,the gray mean value,the position and the area,the feature parameter building model of parts is validated. Experimental results show that this algorithm can detect the complex and various defects of the starterwith low cost,high detection rate and low false detection rate,which has good practical value for the appearance detection of industrial parts.
出处
《电子测量与仪器学报》
CSCD
北大核心
2018年第2期1-8,共8页
Journal of Electronic Measurement and Instrumentation
基金
国家自然科学基金(61605099)
北京市未来芯片技术高精尖创新中心科研基金(KYJJ2016002)
“青年人才托举工程”2016—2018年度项目(2016QNRC001)
关键词
机器视觉
启动器
外观缺陷
缺陷检测
machine vision
starter
appearance defects
defect detection
