摘要
为了解决复杂外形结构的汽车涡轮壳零件表面质量检测难题,研究了基于机器视觉的表面缺陷检测方法。首先,分别采用高角度和低角度打光方式,获取凹坑类和划痕类表面缺陷图像,并通过图像滤波和图像增强提高图像的信噪比。然后,采用基于连通域的轮廓提取算法,确定感兴趣区域。最后,在感兴趣区域内利用灰度分割、形态学处理和断点连接等算法分割出缺陷特征,从而得到零件表面缺陷检测结果。选取60个涡轮壳缺陷件进行了测试。测试结果表明,该方法能够检测出直径0.5 mm以上的凹坑类和划痕类缺陷,漏检率低于1%,单工位检测时间小于2 s。基于机器视觉的表面缺陷检测方法能够替代人工完成复杂结构零件的表面质量检测任务,并有效提高了检测精度和检测效率。实际应用中,该方法配合机械臂还可进一步提高检测系统的柔性。
In order to solve the problem of surface quality inspection of automobile turbine housing with complex shape structure,the method of surface defect inspection based on machine vision is studied.Firstly,the surface defect images of pits and scratches are obtained by high angle and low angle polishing respectively,and the signal-to-noise ratio of the image is improved by image filtering and image enhancement.Then,the contour extraction algorithm based on connected region is used to determine the region of interest.Finally,in the region of interest,gray segmentation,morphological processing and breakpoint connection are used to segment the defect features,so as to get the surface defect detection results.60 defective parts of turbine housing were tested.The test results show that the method can detect pit and scratch defects with a diameter of more than 0.5 mm.The rate of missing detection is less than 1%,and the detection time of single station is less than 2 seconds.The surface defect detection method based on machine vision can replace human to complete the surface quality detection task of complex structural parts,and effectively improve the detection accuracy and efficiency.In practical application,the cooperation with the manipulator can further improve the flexibility of the detection system.
作者
余厚云
张辉
YU Houyun;ZHANG Hui(College of Mechanical and Electrical Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China;Wuxi Institute,Nanjing University of Aeronautics and Astronautics,Wuxi 214187,China)
出处
《自动化仪表》
CAS
2020年第11期6-10,共5页
Process Automation Instrumentation
基金
国家自然科学基金资助项目(51975293)。
关键词
表面质量
缺陷检测
涡轮壳零件
机器视觉
感兴趣区域
断点连接
Surface quality
Defect detection
Turbine housing
Machine vision
Region of interest
Breakpoint connection