光学 精密工程, 2018, 26 (9): 2190, 网络出版: 2018-12-16
BGA焊球视觉检测算法及系统设计
Design of vision detection algorithm and system for BGA welding balls
机器视觉 缺陷检测 球栅阵列 圆度 锡球 分类器 computer vision defect detection Ball Grating Array(BGA) roundness solder ball classifier
摘要
为了实现对BGA焊球的自动检测, 建立了自动视觉检测系统。对系统所采用的焊球特征进行提取及缺陷识别, 基于高斯混合模型的分类器对检测算法进行研究。根据焊球的形状和尺寸特征设计了焊球缺陷识别和分类算法, 并以锡多、锡少和毛刺缺陷为例, 分析典型缺陷的识别算法。以焊球形状的圆度和特征区域的面积等特征参数为评价标准, 构建二维特征空间。在二维特征空间线性组合的基础上, 构建基于高斯混合模型的分类器。构建了训练样本集, 并对该分类器进行训练, 根据训练结果并结合应用实际修正了模型, 并采用测试集对该分类器进行测试验证。实验结果表明, 焊球缺陷检测算法的准确度为97.06%, 漏判率为0%, 检测可靠度为100%。该视觉检测系统满足了工程运用中对识别准确度、稳定性、可靠性等方面的要求。
Abstract
A vision inspection system was proposed for realizing automated detection of Ball Grating Array (BGA) solder balls. An algorithm for solder ball feature extraction, defect recognition, and classification based on the Gaussian mixture model was developed. The recognition and classification algorithm analyzed weld ball defects according to the shape and size characteristics of the welding ball, and the identification algorithm analyzed typical defects by considering excessive solder, solder deficiency, and burr defects as example cases. The two-dimensional feature space was constructed by considering characteristic parameters such as the standard of solder balls shape and feature area as evaluation criteria. Based on the linear combination of two-dimensional feature space, a classifier incorporating the Gaussian mixture model was designed. A sample dataset was employed for training the classifier, and the proposed model was thereby modified according to the obtained training results and production practices. The proposed classifier was evaluated by constructing a test dataset. The obtained experimental results show that the accuracy of the solder ball defect detection algorithm is 97.06%, the leak detection rate is zero, and the detection reliability is 100%. Hence, the proposed model can meet the requirements of recognition accuracy, stability, and reliability needed for realizing an automated visual inspection system.
罗志伟, 杨玉龙, 李志红. BGA焊球视觉检测算法及系统设计[J]. 光学 精密工程, 2018, 26(9): 2190. LUO Zhi-wei, YANG Yu-long, LI Zhi-hong. Design of vision detection algorithm and system for BGA welding balls[J]. Optics and Precision Engineering, 2018, 26(9): 2190.