计算机应用 ›› 2009, Vol. 29 ›› Issue (12): 3375-3377.

• 图形图像处理 • 上一篇    下一篇

基于视觉的SMT印刷钢网尺寸测量方法

张舞杰1,李迪1,叶峰2   

  1. 1. 华南理工大学
    2.
  • 收稿日期:2009-06-03 修回日期:2009-08-07 发布日期:2009-12-10 出版日期:2009-12-01
  • 通讯作者: 张舞杰
  • 基金资助:
    面向电子制造的全自动精密影像测量关键技术及装备

Vision based dimension measurement method for SMT stencil

  • Received:2009-06-03 Revised:2009-08-07 Online:2009-12-10 Published:2009-12-01

摘要: 为实现SMT钢网尺寸的精确自动测量,提出了一种基于Gerber文件信息的视觉测量方法。首先,采用NED公司的NUCLi 7K线扫描相机采集真实图像,并通过读取Gerber文件提取图形的形状及大小等信息。其次,通过坐标比例、旋转和平移变换建立Gerber文件图形信息坐标系和真实图像坐标系之间的坐标映射以确定图形在真实图像上的大致位置。然后,采用Canny算子和Gray moments算子实现图形边缘的像素级和亚像素级精确定位。最后,根据图形的亚像素边缘通过直线和圆等图元拟合方法实现图形尺寸的精确测量。实验中测量值的最大测量误差为0.0854pixel(0.91μm),均方差最大值为0.0282pixel(0.30μm),表明该测量方法具有稳定、可靠和精度高等特点,能有效满足钢网尺寸自动测量要求。

关键词: SMT印刷钢网, 坐标映射, Gerber文件, 亚像素边缘定位, 图形拟合

Abstract: For realizing automatic and precise dimension measurement of Surface Mounting Technology (SMT) stencil, a visual measurement method was proposed based on Gerber file information. First, the real image was grabbed by line scanning camera of NUCLi 7K made by NED company and such information as size and shape was derived from Gerber file data, and the coordinates mapping between figure information coordinates system and the real image coordinates system was erected through coordinates transform of proportion, transportation and rotation for obtaining the approximate position of figures in the real image. Then Canny operator and Gray moment operator were adopted to get the pixel and sub-pixel precision edge. Finally figure fitting methods like line and circle fitting were used to fit figures in the real image for measuring figure dimension precisely. The max measure error is 0.0854pixel (0.91μm) and the max STDEV is 0.0282pixel (0.30μm) in the experiment, which show that: the stencil measurement method proposed in this paper is stable, reliable and of high precision, which can meet the requirements of automatic dimension measurement of SMT stencil efficiently.

Key words: Surface Mounting Technology (SMT) stencil, coordinates mapping, Gerber file, sub-pixel edge location, figure fitting