基于机器视觉的SMT芯片检测方法分析-光学工程专业论文.docxVIP

摘要 I 摘要 电子产品发展日趋精密，芯片检测技术在半导体行业中十分重要。本文主要 研究利用机器视觉对表面贴片式封装芯片的检测方法，实现了针对 SOP 双排式 芯片的管脚识别定位和外观尺寸检测。 文章分析设计了和检测方法相配套的机器视觉软硬件系统，包括光源、相机、 镜头等选取和软件界面的实现。实现了基于图像形态学处理的管脚定位方法，首 先运用顶帽变换处理待测芯片图像，解决了由于光照不均造成的“管脚缺失”现 象；然后利用连通域像素标记方法定位管脚，统计出芯片管脚个数及其几何特征 参数。实现了基于边缘检测的芯片外观尺寸检测方法，首先对待测图像进行直方 图均衡化预处理；然后分析对比经典边缘检测算子效果；优化了滤波算法以解决 中值、均值滤波算法不能完全消除背景噪声的问题；最后利用垂直灰度投影方法 计算管脚宽度和管脚间距，并分析判断芯片管脚缺陷情况。 运用 vc6.0 软件开发平台实现上述检测方法，进行仿真实验。分析实验数据， 文章实现的检测方法能够满足实际检测需求。 关键字：机器视觉 SOP 芯片 边缘检测 连通域像素标记 Ab Abstract III Abstract Electronic products are increasingly sophisticated, chip test is inportent in the semiconductor industry. This paper mainly studies the detection methods of the SMT chip using machine vision, to achieve locating pins and dimensions detection for SOP chip. The paper designs hardware and software system of machine vision matching with detection methods, including light, camera, lens and the interface of software. Achieving the pin positioning method based on morphology, firstly solve the “pin missing” problem of image due to uneven illumination using the top-hat transform; and then locate pins of chip using pixel labeled algorithm of connecting area,counting the number and geometric parameters of pins. Achieving the method of inspection for dimension of chip based on edge detection, firstly using histogram equalization; and then compare the effect of classic edge detection operator; optimizing the filtering algorithm to solve the problem that median, mean filtering algorithm can not completely eliminate background noise; lastly calculate the width and pitch of pins using vertical gray projection and analyse defect of pins. Achieving the testing methods above mentioned using vc6.0 software development platform. According to the experimental data, the methods mentioned in the article could meet the needs of the actual test Keywords: machine vision sop chip edge detect pixel labeled in connecting area 目录 目录 Ⅴ 基于机器 基于机器视觉的 SMT 芯片检测研究 VI 4.5 芯片管脚尺寸检测36 4.6 本章小结39 第五章 总结与展望41