基于机器视觉的计量检定流水线非标异物入侵自动化识别方法

Automatic Identification Method for Non-standard Foreign Object Intrusion in Metrological Verification Pipeline Based on Machine Vision

在线阅读下载PDF

导出

摘要非标准型的异物种类和尺寸差异性较大,识别过程较慢,识别结果准确率较低,导致流水线异物卡顿现象频发。对此,以计量检定流水线为研究对象,提出了一种基于机器视觉技术的计量检定流水线非标异物入侵自动化识别方法。获取不同区域之间的颜色和纹理特征,计算卡方距离,实现异物空间测量,分析世界坐标系和相机坐标系、相机坐标系与图像坐标系之间的齐次坐标变换过程,对异物中心点坐标进行转换,采用基于Hough变换的直线检测方法,通过边缘算子提取直线特征值,实现物体表面定位,校正非标异物图像产生的畸变,利用帧差法进行多图像差分,结合多角度成像,并计算四图像之间差值,计算出异物中心坐标,实现异物入侵识别。研究结果表明,不同种类异物入侵下,设计的基于机器视觉的计量检定流水线非标异物入侵自动化识别方法误差能够控制在5%以内,验证了方法有效性。 The types and sizes of non-standard foreign objects vary greatly,and the recognition process is slow,resulting in low accuracy of recognition results and frequent occurrence of foreign object jamming in the assembly line.In this regard,a machine vision based automated identification method for non-standard foreign object intrusion in the metrological verification assembly line is proposed,taking the metrological verification assembly line as the research object.Obtain color and texture features between different regions,calculate chi square distance,achieve foreign object spatial measurement,analyze the homogeneous coordinate transformation process between world coordinate system and camera coordinate system,camera coordinate system and image coordinate system,convert the center point coordinates of foreign objects,use line detection method based on Hough transform,extract line feature values through edge operator,and achieve object surface positioning,correct the distortion caused by non-standard foreign object images,use frame difference method for multi image differentiation,combine with multi angle imaging,and calculate the difference between the four images to calculate the center coordinates of foreign objects,achieving foreign object intrusion recognition.The research results indicate that under different types of foreign object intrusion,the error of the designed machine vision based automated identification method for non-standard foreign object intrusion in the metrological verification pipeline can be controlled within 5%,which verifies the effectiveness of the method.

作者鲁观娜李亮刘影袁瑞铭王慧楠 LU Guanna;LI Liang;LIU Ying;YUAN Ruiming;WANG Huinan(Center of Metrology,State Grid Jibei Electric Power Company Limited,Beijing 102208,China)

机构地区国网冀北电力有限公司计量中心

出处《计算技术与自动化》 2025年第2期142-146,共5页 Computing Technology and Automation

关键词机器视觉计量检定非标异物入侵自动化识别坐标系转换 machine vision metrological verification non-standard foreign objects intrusion automatic identification coordinate system conversion

分类号 TP274.5 [自动化与计算机技术—检测技术与自动化装置]

作者简介通信联系人:鲁观娜(1981-),女,辽宁沈阳人,硕士,高级工程师,研究方向:电气测量技术和管理。E-mail:lianjiao417815@163.com。