【作者】 章坚；

【导师】 陈隆道； 张利华；

【作者基本信息】 浙江大学 ， 电气工程， 2012， 硕士

【摘要】 印刷电路板(PCB, Printed Circuit Board)是集成各种电子元器件的信息载体,在各个领域得到了广泛的应用,并且伴随着生产技术的不断提高,PCB板朝着多层数,高密度的方向发展。由于PCB制作工艺日趋复杂,使PCB的质量检验成为一件非常困难的工作。传统的检测方法已不能满足生产的需要。基于计算机与图像处理技术进行的PCB自动视觉检测的研究开始成为PCB检测的热门方向。本文主要研究利用目前工业中应用最多的BP (BP,Back Propogation)神经网络和AOI(Automatic Optic Inspection,自动光学检测)技术,借助计算机视觉,根据电表PCB的元件特征,进行PCB板图像的匹配识别。本文主要就以下几个问题进行了具体研究首先介绍了研究自动检测PCB板对于行业的重要性及采用的技术手段,PCB板图像识别的基本思想和技术特点；随后,阐述了PCB板元件图像的特征提取方式,介绍了图像的处理相关的算法；然后,本文详细描述了神经网络在该系统中的实现和技术特点。设计和制作了相应的检测设备,编写了配套的操作软件用于采集图像,匹配识别,分析结论,给出报表。实验结果表明本方法误判率低、匹配精度高,可以应用于PCB板中元件的一致性检测。更多还原

【Abstract】 Print Circuit Board (PCB) is an information carrier integrating various electronic components. It has been applied in different fields widely.also with the continuous development of production technology, PCB are prone to be more layers and more density。Because of the complexity of PCB production process, makes the PCB quality inspection become a very difficult job. The traditional examination method can not satisfy the production’s needs already. The study on the automatic vision examination of PCB flaw is popular in the PCB examination, which is based on the computer and the image processing technology.This paper mainly research on the application of BP(Back Propogation) neural network and AOI (Automatic Optic Inspection) Technology which most widely used in industry, with the help of computer vision, according to characteristics of the meter PCB components, carrying out the matching recognition of the PCB’s image. This review focuses on the following specific research issues.First, it is introduced the importance of Automatic detection of PCB board and the technical means used, the basic idea and technical features of PCB board image recognition; next, it is illustrated the extraction method of the PCB board components image’s feature, and introduced the related algorithms for image processing; then,this paper described the implementation and technical features of neural network in this systems in detailed. We have designed and produced the corresponding detection device, prepared the supporting operating software for image acquisition, matching recognition, result analysis, reports given. The experimental results indicates that this means have low error decision rate, high matching precision, can be used in testing the consistency of PCB board’s components.更多还原