基于图像特征统计分析的PCB焊点检测方法.pdfVIP

学兔兔 第32卷 第2期 仪 器 仪 表 学 报 V01．32 NO．2 2011年2月 Chinese Journal of Scientific Instrument Feb．20l1 基于图像特征统计分析的PCB焊点检测方法 谢宏威，张宪民，邝泳聪 (华南理工大学机械与汽车工程学院 广州 5 10640) 摘 要：提出了一种基于图像特征统计分析的炉后焊点检测方法，以提高在线自动光学检测系统的检测性能和可操作性。提 出双阈值的AdaBoost算法用于设计分类器，在训练的同时进行最优特征选择和分类器的增强，实现了焊点图像特征的自动提 取和检测参数的自动设定。采用分类和回归树方法将焊点缺陷决策方法优化为一棵二叉决策树，提高了检测速度。实验结果 表明，该方法训练速度较快，可以满足实际生产需要。与目前已经实用化的罔像对比算法和图像分析算法相比，在保持现有检 测速度基本不变的情况下，该方法的检测精度更高。 关键词：自动光学检测；特征选择；缺陷决策；统计学习 中图分类号：TP391 文献标识码：A 国家标准学科分类代码：520．604 Solder joint inspection method based on image feature statistical analysis Xie Hongwei，Zhang Xianmin，Kuang Yongcong (School of Mechanical and Automotive Engineering，South China University of Technology，Guangzhou 510640，China) Abstract：In order to improve the performance and maneuverability of online automatic optical inspection system，a novel method for post—reflow solder joint inspection based on image feature statistical analysis is proposed in this pa— per．Dual—threshold AdaBoost algorithm is used to design the classifier．Optimal features are selected and classifier is enhanced during the training process．Features of the solder joints are selected and detection parameters are set auto— matically．The solder joint defect decision method is transformed into a binary decision tree to improve the detecting speed．Experimental resuhs prove that the training time is short enough to satisfy the actual production requirements． Compared with the image comparison method and image feature analysis method that have been applied in practical production，the method proposed in this paper has higher inspection precision while maintaining detecting speed． Key