基于计算机视觉的PCB板焊点缺陷检测及分类
发布时间：2018-11-29 12:59
【摘要】：印刷电路板(Printed Circuit Board,PCB)具有高密度、高精度、高可靠性的特点;并且随着PCB板的层数增加,密度更高,PCB板的体积也越来越小,使得PCB板焊点缺陷的检测更加困难。本文结合格力公司PCB板特定区域焊点缺陷检测与识别的背景需求,基于计算机视觉技术,对PCB板焊点缺陷进行分析,针对短路,漏焊,少锡,多锡等四种典型缺陷的检测及识别,提供实用的解决方案。论文工作及主要贡献如下:(1)PCB板图像的颜色通道的选择。PCB板焊点区域在不同颜色空间、不同颜色通道呈现不同的图像特性。为确保PCB板焊点区域缺陷的有效检测与识别,本文首先进行了PCB板焊点图像的颜色空间及颜色通道的选择。实验表明基于RGB颜色空间下的R通道的PCB板焊点图像描述更为合适。(2)基于形态学的PCB板焊点缺陷检测及识别。针对PCB板焊接图像的R通道,首先借助滤波抑制噪声干扰;在此基础上,进行图像的二值化,借助连通区域标记以及二值图像的形态学滤波,获取感兴趣焊点区域的相关特征;构造焊点缺陷检测与识别规则,实现焊点缺陷的检测,并识别出相应的焊点缺陷类型。由于焊点缺陷的识别规则中涉及焊点区域面积的绝对阈值,当监控摄像头与检测流水线的相对位置发生改变时,这些面积阈值需要重新选择。(3)基于图像配准的PCB板焊点缺陷检测及识别。该方法首先借助摄像头获取未经焊接的PCB板图像,作为标准参考图像;选取PCB板中待检测的焊点位置及感兴趣区域;借助自动图像配准,将检测流水线上的PCB板焊接图像中感兴趣的焊点区域进行几何校正;以参考图像中焊点模板为基准,构建焊点缺陷检测与识别规则,根据焊点特征进行缺陷检测及焊点缺陷类型的识别。当监控摄像头关于检测流水线的相对位置发生改变,只需重新获取标准参考图像即可,而无须进行有关阈值参数的重新设定。实验结果表明,基于图像配准的PCB焊点缺陷检测方法,具有一定的实时性,可满足相应产品的生产需求。
[Abstract]:Printed circuit board (Printed Circuit Board,PCB) has the characteristics of high density, high precision and high reliability, and with the increase of the number of layers of PCB, the density is higher and the volume of PCB board is smaller and smaller, which makes the detection of solder joint defects of PCB plate more difficult. In this paper, based on computer vision technology, the defects of PCB plate solder joint defects are analyzed based on the background requirements of Gree company PCB plate specific area solder joint defect detection and identification, aiming at short circuit, missing welding, less tin, The detection and identification of four typical defects, such as polytin, provide practical solutions. The main contributions of this paper are as follows: (1) the selection of color channels for PCB board images. The different color channels of PCB plates show different image characteristics in different color spaces. In order to ensure the effective detection and recognition of the defects in the solder joint area of PCB plate, the color space and color channel of the PCB solder joint image are firstly selected in this paper. Experiments show that the image description of PCB solder joint based on R channel in RGB color space is more suitable. (2) defect detection and recognition of PCB plate solder joint based on morphology. Aiming at R channel of PCB welding image, the noise interference is suppressed by filtering, then the binarization of image and the morphological filtering of connected region and binary image are carried out to obtain the relevant features of solder joint region of interest. The detection and identification rules of solder joint defects are constructed to realize the detection of solder joint defects and the corresponding types of solder joint defects are identified. Because the absolute threshold of solder joint area is involved in the recognition rules of solder joint defects, when the relative position of the surveillance camera and the detection pipeline changes, These area thresholds need to be re-selected. (3) PCB plate solder joint defect detection and recognition based on image registration. The method firstly uses the camera to obtain the image of the unwelded PCB board as the standard reference image, and selects the spot position and the region of interest of the PCB board to be detected. With the help of automatic image registration, geometric correction of the solder joint area of interest in the PCB plate welding image on the detection pipeline is carried out. Based on the reference pattern of solder joint in the image, the rules of detection and recognition of solder joint defects are constructed, and the defect detection and type identification are carried out according to the characteristics of solder joint. When the relative position of the surveillance camera about the detection pipeline changes, it is only necessary to obtain the standard reference image again, without the need to reset the threshold parameters. The experimental results show that the PCB solder joint defect detection method based on image registration has a certain real-time performance and can meet the production requirements of the corresponding products.
【学位授予单位】：河北师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN41;TP391.41
本文编号：2365059
[Abstract]:Printed circuit board (Printed Circuit Board,PCB) has the characteristics of high density, high precision and high reliability, and with the increase of the number of layers of PCB, the density is higher and the volume of PCB board is smaller and smaller, which makes the detection of solder joint defects of PCB plate more difficult. In this paper, based on computer vision technology, the defects of PCB plate solder joint defects are analyzed based on the background requirements of Gree company PCB plate specific area solder joint defect detection and identification, aiming at short circuit, missing welding, less tin, The detection and identification of four typical defects, such as polytin, provide practical solutions. The main contributions of this paper are as follows: (1) the selection of color channels for PCB board images. The different color channels of PCB plates show different image characteristics in different color spaces. In order to ensure the effective detection and recognition of the defects in the solder joint area of PCB plate, the color space and color channel of the PCB solder joint image are firstly selected in this paper. Experiments show that the image description of PCB solder joint based on R channel in RGB color space is more suitable. (2) defect detection and recognition of PCB plate solder joint based on morphology. Aiming at R channel of PCB welding image, the noise interference is suppressed by filtering, then the binarization of image and the morphological filtering of connected region and binary image are carried out to obtain the relevant features of solder joint region of interest. The detection and identification rules of solder joint defects are constructed to realize the detection of solder joint defects and the corresponding types of solder joint defects are identified. Because the absolute threshold of solder joint area is involved in the recognition rules of solder joint defects, when the relative position of the surveillance camera and the detection pipeline changes, These area thresholds need to be re-selected. (3) PCB plate solder joint defect detection and recognition based on image registration. The method firstly uses the camera to obtain the image of the unwelded PCB board as the standard reference image, and selects the spot position and the region of interest of the PCB board to be detected. With the help of automatic image registration, geometric correction of the solder joint area of interest in the PCB plate welding image on the detection pipeline is carried out. Based on the reference pattern of solder joint in the image, the rules of detection and recognition of solder joint defects are constructed, and the defect detection and type identification are carried out according to the characteristics of solder joint. When the relative position of the surveillance camera about the detection pipeline changes, it is only necessary to obtain the standard reference image again, without the need to reset the threshold parameters. The experimental results show that the PCB solder joint defect detection method based on image registration has a certain real-time performance and can meet the production requirements of the corresponding products.
【学位授予单位】：河北师范大学
【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4条
1 白木,周洁;AOI系统在SMT生产线上的应用[J];电子机械工程;2003年03期
2 孙亦南,刘伟军,王越超,孙亦蓬;一种用于圆检测的改进Hough变换方法[J];计算机工程与应用;2003年20期
3 李锐;王卷乐;郭复胜;杜锦华;;Fourier-Mellin变换不同时相遥感影像自动配准研究[J];计算机工程与应用;2010年16期
4 李晓明;赵训坡;郑链;胡占义;;基于Fourier-Mellin变换的图像配准方法及应用拓展[J];计算机学报;2006年03期
相关硕士学位论文 前2条
1 李薇;基于图像处理的印刷电路板缺陷检测技术研究[D];重庆大学;2007年
2 刘阳;基于图像处理的PCB焊接缺陷检测技术研究[D];大连理工大学;2009年
，本文编号：2365059
本文链接：https://www.wllwen.com/kejilunwen/dianzigongchenglunwen/2365059.html
