摘要
针对陶瓷阀芯表面缺陷特征复杂、啮合表面存在研磨抛光区域等问题,现有基于视觉检测的陶瓷阀芯表面缺陷检测算法对不同种类陶瓷阀芯的适应性较差且漏检率高,该研究提出了一种分区域、多级优化的陶瓷阀芯表面缺陷检测算法。首先依据陶瓷阀芯表面反射率的不同,将其划分为两个检测区域:抛光面(P)、非抛光面(Non P),然后根据陶瓷阀芯表面缺陷特征,将其分为两类缺陷:G缺陷、C缺陷。在检测过程中,搜索模板图像金字塔信息,调整目标灰度图像位姿,经掩模板(Mask)配准,提取出P区域、Non P区域,通过blob检测、裂纹检测,分别对P区域、Non P区域进行缺陷提取。通过实验研究与系统设计,分析了检测算法的可行性、适用性,结果表明该研究算法针对不同种类陶瓷阀芯表面缺陷检测漏检率低,且能通过优先级的合理设置,提高检测效率。
In order to deal w ith the problems such as the ceramic valve core surface defects are complex and existing polishing region in grinding surface. There are poor adaptability and high miss rate for different types of ceramic valve core based on visual inspection by spool surface defect detection algorithm. A novel defect detection algorithm w ith multistage optimization w as proposed. According to the different reflectivity in ceramic spool surface,the ceramic spool surface w as divided into tw o regions: polished surface( P) and non-polished surface( Non P),meanw hile by the characteristics of ceramic valve core surface defect,it w as divided into tw o types of defect: G defects and C defects. In the detection process,search Template Image Pyramid Information,adjust the target grayscale image,masked by mask,extract the P region,Non P area,use blob detection,crack detection,respectively,the P area,Non P area for defect extraction. Through the experimental study and system implementation,the feasibility and applicability of the detection algorithm is analyzed. The results show that: the miss rate of the algorithm is low for the different types of ceramic spool,and can improve the detection efficiency by setting reasonable priority.
出处
《组合机床与自动化加工技术》
北大核心
2017年第10期82-86,90,共6页
Modular Machine Tool & Automatic Manufacturing Technique
基金
湖北省教育厅重点项目:面向芯片贴装的显微视觉在线检测技术研究(D 20151406)
湖北省自然科学基金项目:多目视觉引导机器人绝对定位技术的研究(2016CFB513)
关键词
陶瓷阀芯
掩模板
blob检测
裂纹检测
ceramic valve core
mask template
blob detection
crack detection
