PCB人工焊接缺陷检测与识别算法研究
发布时间：2019-01-19 14:21
【摘要】：在现如今的电子工业中,印制电路板(PCB)作为电子元器件的载体,其板载元器件的焊接质量将直接影响到电子产品的性能,因此对PCB焊接质量的检测对工业生产而言意义重大。传统检测手段耗时耗力且可靠性差,为了提高工业生产效率、改善电子产品的性能和质量,降低生产所耗成本,采用基于图像处理技术的自动化焊接缺陷检测方法,可以达到非接触且高精度的检测效果。 课题研究PCB人工焊接缺陷检测与识别算法。实验选用普通人工焊接的PCB为研究对象,经扫描及焊点配准后获得PCB焊点图像。对焊点图像进行灰度处理、中值滤波等预处理操作,以减少图像噪声,提高焊点图像的清晰度。本文研究焊点图像特征的提取方法,提出基于阈值分割技术提取焊点图像的形状特征；根据阈值分割结果提取焊点图像前景和背景特征点,采用支持向量机(SVM)方法对原彩色焊点图像进行分类后并做灰度处理,得到只包含前景图像的焊点灰度图像,提取焊点灰度图像的小波特征。 在完成焊点图像的特征提取后,针对缺陷识别方法进行研究,提出一种基于特征聚集度的模糊C均值聚类(FCM)与松弛约束支持向量机(RSVM)联用的分类识别算法。算法首先对样本特征数据进行模糊C均值聚类,依据样本隶属度函数计算不同特征的特征聚集度,并由特征聚集度指标改进RSVM算法中的松弛量参数,建立最终的分类器模型。实验表明,本文提出的算法建立了泛化能力更强的分类模型,能有效抑制噪声及模糊边界点对分类模型的影响,在人工焊接缺陷识别的应用中获得了满意的识别结果。
[Abstract]:In today's electronic industry, printed circuit board (PCB), as the carrier of electronic components, the welding quality of board components will directly affect the performance of electronic products. Therefore, the detection of PCB welding quality is of great significance to industrial production. In order to improve the efficiency of industrial production, improve the performance and quality of electronic products and reduce the cost of production, the automatic welding defect detection method based on image processing technology is adopted. Can achieve non-contact and high-precision detection effect. The PCB artificial welding defect detection and identification algorithm is studied in this paper. The PCB of ordinary manual welding was selected as the research object, and the PCB solder joint image was obtained after scanning and solder joint registration. In order to reduce the image noise and improve the definition of solder joint image, the gray level processing and median filtering are used to preprocess the solder joint image. In this paper, the feature extraction method of solder joint image is studied, and the shape feature of solder joint image is extracted based on threshold segmentation technique. According to the threshold segmentation results, the solder joint image foreground and background feature points are extracted, and the original color solder joint image is classified by support vector machine (SVM) method, and the gray level image containing only foreground image is obtained. Wavelet feature of solder joint gray image is extracted. After the feature extraction of solder joint image is completed, the defect recognition method is studied, and a fuzzy C-means clustering (FCM) based on feature aggregation and relaxation constraint support vector machine (RSVM) is proposed. Firstly, the fuzzy C-means clustering of the sample feature data is carried out, and the characteristic aggregation degree of different features is calculated according to the membership function of the sample. The relaxation parameter in RSVM algorithm is improved by the index of feature aggregation degree, and the final classifier model is established. The experimental results show that the proposed algorithm can effectively suppress the influence of noise and fuzzy boundary points on the classification model and obtain satisfactory recognition results in the application of manual welding defect recognition.
【学位授予单位】：华东理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN41;TP391.41
本文编号：2411453
[Abstract]:In today's electronic industry, printed circuit board (PCB), as the carrier of electronic components, the welding quality of board components will directly affect the performance of electronic products. Therefore, the detection of PCB welding quality is of great significance to industrial production. In order to improve the efficiency of industrial production, improve the performance and quality of electronic products and reduce the cost of production, the automatic welding defect detection method based on image processing technology is adopted. Can achieve non-contact and high-precision detection effect. The PCB artificial welding defect detection and identification algorithm is studied in this paper. The PCB of ordinary manual welding was selected as the research object, and the PCB solder joint image was obtained after scanning and solder joint registration. In order to reduce the image noise and improve the definition of solder joint image, the gray level processing and median filtering are used to preprocess the solder joint image. In this paper, the feature extraction method of solder joint image is studied, and the shape feature of solder joint image is extracted based on threshold segmentation technique. According to the threshold segmentation results, the solder joint image foreground and background feature points are extracted, and the original color solder joint image is classified by support vector machine (SVM) method, and the gray level image containing only foreground image is obtained. Wavelet feature of solder joint gray image is extracted. After the feature extraction of solder joint image is completed, the defect recognition method is studied, and a fuzzy C-means clustering (FCM) based on feature aggregation and relaxation constraint support vector machine (RSVM) is proposed. Firstly, the fuzzy C-means clustering of the sample feature data is carried out, and the characteristic aggregation degree of different features is calculated according to the membership function of the sample. The relaxation parameter in RSVM algorithm is improved by the index of feature aggregation degree, and the final classifier model is established. The experimental results show that the proposed algorithm can effectively suppress the influence of noise and fuzzy boundary points on the classification model and obtain satisfactory recognition results in the application of manual welding defect recognition.
【学位授予单位】：华东理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN41;TP391.41
【参考文献】
相关期刊论文 前10条
1 周德俭;SMT组装质量检测中的AOI技术与系统[J];电子工业专用设备;2002年02期
2 姚立新,张武学,连军莉;AOI系统在PCB中的应用[J];电子工业专用设备;2004年05期
3 迟学芬,韩昌元,金辉;CCD推扫遥感成像中的欠采样噪声处理[J];电子学报;2003年09期
4 孙劲光,张文斌,朱世安;图象灰度的处理方法及实现[J];辽宁工程技术大学学报(自然科学版);2002年03期
5 鲜飞;;AOI技术在SMT生产上的实施技巧与策略[J];国内外机电一体化技术;2006年01期
6 李薇;;PCB自动光学检测系统[J];工业控制计算机;2009年01期
7 李捷;唐星科;蒋延军;;几种边缘检测算法的比较研究[J];信息技术;2007年09期
8 卢盛林;张宪民;邝泳聪;;基于神经网络的PCB焊点检测方法[J];华南理工大学学报(自然科学版);2008年05期
9 万永菁;张佩;钱佳;;一种融合图像滤波的FCM图像分割算法及其应用[J];华东理工大学学报(自然科学版);2013年02期
10 吴平峰;代宣军;;PCB测试技术的介绍[J];现代机械;2009年04期
，本文编号：2411453
本文链接：https://www.wllwen.com/kejilunwen/dianzigongchenglunwen/2411453.html
