【作者】 杨志高；

【导师】 谢光汉；

【作者基本信息】 广东工业大学 ， 控制理论与控制工程， 2005， 硕士

【摘要】 集成电路(IC)是现代信息产业和信息社会的基础,IC产业已成为国民经济和国防建设的基础性、战略性产业。目前,中国的IC设计和制造水平同世界先进国家相比存在着很大的差距。因此,IC制造设备的开发对中国集成电路产业的发展具有十分重要的意义。 引线键合技术是现代微电子封装中常用的一种技术。随着芯片集成度的增加和尺寸的减小、芯片引脚间距越来越细密化,微电子封装的精度要求越来越高。引线键合工艺过程的每个步骤都涉及到定位问题,以往这种带有高重复性和智能性的工作只能由人眼来完成,而有些时候对微小尺寸的精确快速测量、形状匹配、颜色辨识等,用人眼根本无法连续稳定地进行。 本文利用机器视觉技术与数字图像处理技术来实现智能IC引线键合机的图像识别系统。本文主要研究内容有: 首先,对IC引线框架的定位设计做了比较全面的研究。针对IC引线框架的左下角和右上角有两个非常明显的圆标志,又由示教系统和位置误差信息,可以推算出圆标志的潜在区域。在每个圆标志的潜在区域里,对采集摄入的数字图像,进行图像预处理、边缘提取得到离散的边缘特征点,然后对其进行改进的霍夫变换得到圆心坐标,从而分别确定两个圆标志的精确位置。 其次,对IC裸芯特征点的定位设计做了深入的研究。一旦引线框架定位了,则IC裸芯特征点的潜在区域也随之确定。在其潜在区域,对图像进行图像增强、平滑滤波、二值化得到二值图像,然后对得到的二值图像进行图像形态处理(先“膨胀”后“腐蚀”),之后采用小连通区域删除方法对干扰小区域的快速删除。在得到几个大的分块区域后,又对区域分块标注进行了深入的研究,能够快速地对每个特征点进行特征提取,从而精确地对IC裸芯进行了定位设计。 最后,本文对键合完成后的键合点进行检测系统的设计。通过图像预处理、键合区分割、各分割区域的形状参数提取,对采集到的键合点的焊区形状、面积等进行分析判断,从而实现全自动检测,从而有效地提高生产效率。 本文研究的图像识别系统对于智能IC引线键合机的研究开发提供了关键技术更多还原

【Abstract】 Integrate circuit (IC) is the foundation of modern information industry and information society. IC industry has become the basic and strategic industry of national economy and national defence construction. At present, the IC design and manufacture level of our country has existed a large gap comparing to the advanced country of the world. Therefore, to exploit IC manufacturing equipment is of great significance to the development of the integrate circuit industry of our country.Wire bonding today is used throughout the microelectronics industry as a means of interconnecting the chips, substrates and output pins. As Ics continue to become more sophisticated and minimum feature sizes keep shrinking, the need for higher accuracy in fine pitch interconnect applications has become more important than ever. The problem of fixing position is involved in every step of lead-bonding technics process. In the past, this work with high repetion and aptitude can only be finished by people’s eyes, but sometimes it is impossible for people to do this work continuously and steadily by their eyes.This paper uses machine vision technology and digital image processing technology to realize the image recognition system of the intelligent IC wire bonder. The primary research content of this paper includes:Firstly, an all-around study has been done on how to locate the IC lead-frame. On the left-down side and right-up side of the IC lead frame, there are two very obvious circularity sign by which the potential region of the circularity sign can be calculated. Capturing the digital image in the potential region of every circularity sign and acquiring the discrete edge feature points by image pretreatment and edge pick-up. Then circular coordinate can be acquired by improved hough-transform and precise position更多还原