基于视觉的环形密封件在线检测  

Vision⁃based Ring Seal Online Detection

作　　者：宋薇[1] 陈兴 沈林勇[1] SONG Wei;CHEN Xing;SHEN Lin⁃yong(School of Mechanical Engineering and Automation,Shanghai University,Shanghai 200444,China)

机构地区：[1]上海大学机电工程与自动化学院,上海200444

出　　处：《仪表技术与传感器》2020年第5期62-67,共6页Instrument Technique and Sensor

摘　　要：为了对生产线上的密封件进行尺寸检测,从而剔除不合格的产品,提出了一种基于机器视觉的自动在线检测系统。检测对象为环形圈和薄型垫片。当密封件在传送带上运动到相机正下方时,采用多传感器多工位的检测方案,反馈给相机自动采集信号,相机采集到图像后通过模板匹配截取兴趣区域、阈值分割、轮廓提取、最小二乘法拟合等图像处理技术,实现了对密封件的尺寸测量。不合格的产品通过气动装置分拣剔除。实验结果表明,环形圈的检测精度为0.1 mm,薄型垫片的检测精度为0.5 mm,检测时间小于0.75 s。该视觉检测系统达到了在线对密封件进行尺寸检测的要求。In order to inspect the seals on the production line and eliminate the unqualified products,an automatic on⁃line detection system based on machine vision was proposed.The detection object was a kind of annular ring and a kind of thin gasket.When the seal moved on the conveyor belt directly below the camera,the multi⁃sensor multi⁃station detection scheme was used to automatically acquire signals to control the camera to capture the image.After the camera captured the image,the image processing techniques such as intercepting the region of interest by template matching,threshold segmentation,contour extraction and least squares fitting were performed to realize dimensional measurements of the seal.Unqualified products were sorted and removed by pneumatic means.The experimental results show that the detection accuracy of the annular ring is 0.1 mm,the detection accuracy of the thin gasket is 0.5 mm and the detection time is less than 0.75 s.The visual inspection system meets the requirement of on⁃line dimension inspection for seals.

关 键 词：机器视觉 尺寸检测 在线检测 环形密封件 图像处理 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]