华南理工大学学报(自然科学版) ›› 2009, Vol. 37 ›› Issue (9): 98-102,116.

• 机械工程 • 上一篇    下一篇

无铅焊点鲁棒定位的灰度积分投影算法

吴福培 张宪民 邝泳聪 欧阳高飞   

  1. 华南理工大学 机械与汽车工程学院, 广东 广州 510640
  • 收稿日期:2008-08-17 修回日期:2008-12-18 出版日期:2009-09-25 发布日期:2009-09-25
  • 通信作者: 吴福培(1980一),男,博士生,主要从事自动光学检测、机器视觉研究. E-mail:wufupei@163.com
  • 作者简介:吴福培(1980一),男,博士生,主要从事自动光学检测、机器视觉研究.
  • 基金资助:

    国家杰出青年科学基金资助项目(50825504);粤港关键领域突破项目(东莞专项200816822);广州市科技攻关项目(2008A010300002)

Robust Positioning Algorithm of Lead-Free Solder Joints Based on Gray-Level Integration Projection

Wu Fu-pei  Zhang Xian-min  Kuang Yong-cong  Ouyang Gao-fei   

  1. School of Mechanical and Automotive Engineering, South China University of Technology, Guangzhou 510640, Guangdong, China
  • Received:2008-08-17 Revised:2008-12-18 Online:2009-09-25 Published:2009-09-25
  • Contact: 吴福培(1980一),男,博士生,主要从事自动光学检测、机器视觉研究. E-mail:wufupei@163.com
  • About author:吴福培(1980一),男,博士生,主要从事自动光学检测、机器视觉研究.
  • Supported by:

    国家杰出青年科学基金资助项目(50825504);粤港关键领域突破项目(东莞专项200816822);广州市科技攻关项目(2008A010300002)

摘要: 针对自动光学检测系统因焊点定位不准确而导致的误判问题,基于灰度积分投影技术提出了一种焊点定位的鲁棒算法.首先对无铅焊点图像进行预处理,利用焊点的颜色特征对图像进行二值化,将焊点从印刷电路板图像中分割出来;然后分别利用焊点特征的水平和垂直灰度积分投影曲线,以焊盘窗内焊点像素的面积最大化为目标,获得焊点的定位坐标,以实现准确的焊点定位.此外,通过引入Blob评价函数区分焊点与噪声,从而有效地减少了噪声的干扰,提高了定位算法的鲁棒性.实验结果表明,采用所提出的算法对变化多样的Chip元件焊点进行定位时,定位误差小于1个像素,具有较高的定位精度.

关键词: 自动光学检测, 无铅焊点, 焊点定位, 灰度积分投影, 鲁棒性

Abstract:

In order to avoid the misjudgement of automatic optical inspection system due to the positioning error of solder joints,a robust positioning algorithm of solder joints is presented based on the gray-level integration projection.In this algorithm,first,solder joints are segmented from the printed circuit board(PCB) image after the binarization of the pretreated image of lead-free solder joints according to the color feature of the joints.Then,with the help of gray-level integration projection curves of solder joint features in both horizontal and vertical directions,the positioning coordinates of the joints are obtained and the accurate positions of the joints can be determined by maximizing the pixel area of solder joints within the solder-land windows. Moreover, by introducing a Blob evaluation function, the solder joint and the noise are differentiated, thus effectively reducing the disturbance of noise and improving the robustness of the algorithm. Experimental results show that the proposed algorithm is of a high positioning precision less than 1 pixel for various solder joints of a Chip component.

Key words: automatic optical inspection, lead-free solder joint, solder joint location, gray-level integration projection, robustness