摘要
目的针对芯片包装载带在生产过程中经常出现的型腔底部和边缘变形、穿孔等缺陷的检测问题,提出一种机器视觉检测方法。方法首先离线准备配准模板及标准模板图像,然后根据模板在生产过程中进行在线检测。在检测过程中由传感器触发采集待检测型腔图像,然后通过模板匹配方法配准模板图像和待检测图像,并进行异或运算检测两图像差异从而定位缺陷。结果实验证明边缘变形检测最大错误率为0.45%,底部变形检测最大错误率为0.50%,穿孔检测最大错误率为0.35%,每帧图像检测平均耗时为0.22 s,满足用户错误率不超过1%和每帧耗时不超过0.5 s的要求。结论该方法能够实时检测芯片载带边缘变形、穿孔等缺陷,有效地实现载带加工生产过程中的质量监控。
An efficient and accurate machine vision detection method was proposed to detect the deformation and perforation at the bottom and edge of the cavity of chip packaging carrier tape.The registration template and standard template images are prepared offline and then detected online during production.During the detection process,the cavity image to be detected is triggered by the sensor,and then the template image and the image to be detected are registered by the template matching method,and the XOR operation is performed to detect the difference between the two images so as to locate the defect.Experiments show that the maximum error rate of edge deformation detection is 0.45%,the maximum error rate of bottom deformation detection is 0.50%,and the maximum error rate of perforation detection is 0.35%.The average detection time of each frame is 0.22 s,which meets the user's requirement that the error rate is less than 1%and the time of each frame is less than 0.5 s.This method can detect the edge deformation and perforation of the chip in real time,and effectively realize the quality monitoring in the process of chip loading processing.
作者
魏鸿磊
蒋志留
徐家恒
孔祥志
商业彤
童强
WEI Hong-lei;JIANG Zhi-liu;XU Jia-heng;KONG Xiang-zhi;SHANG Ye-tong;TONG Qiang(School of Mechanical Engineering and Automation,Dalian Polytechnic University,Liaoning Dalian 116034,China)
出处
《包装工程》
CAS
北大核心
2022年第11期183-188,共6页
Packaging Engineering
基金
辽宁省教育厅2021年度科学研究经费面上项目(LJKZ0535,LJKZ0526)
2021年度本科教育教学综合改革项目(JGLX2021020,JCLX2021008)。
关键词
芯片载带
缺陷检测
机器视觉
图像异或运算
chip carrier
defect detection
machine vision
image XOR operation
