导　　师： 张宪民;范彦斌

学科专业： 080201

授予学位： 硕士

作　　者： ;

机构地区： 华南理工大学

摘　　要： 随着电子产品朝着高密度和高性能方向发展,对电子产品表面质量检测提出了更高的要求。基于图像识别的自动光学检测/(Automatic Optical Inspection, AOI/)可对高密度贴装PCB/(Printed Circuit Board/)中的缺陷进行可靠的快速检测。本文以电路板板载元器件缺陷为研究对象,结合编程软件VC++和图像处理软件HALCON8.0,对表面贴装电子组件质量检测进行了研究。 首先对AOI系统的工作原理以及硬件系统的组成进行了阐述,系统硬件主要包括光源、相机、镜头、伺服电机和运动控制卡。再对图像处理软件HALCON8.0和VC++编程平台进行了简单介绍。 其次针对相机拍摄视野的限制,利用基于Harris角点特征的图像拼接方法,解决每次拍摄得到只能PCB一个局部图像的问题。该图像拼接算法包括图像特征点检测、特征点匹配和图像融合三个步骤,根据系统移动平台控制精度较高的特点,提出了一种改进的图像融合方法。实验表明,该方法能减少图像拼接所需的时间,而且拼接后的全局图能够作为系统检测时的导航图。 然后针对AOI系统检测过程中PCB出现的平移及旋转情况,利用仿射变换对图像进行对准。研究了本系统所涉及的图像匹配算法,将图像拼接和图像匹配置于不同的线程中,提出了将多线程技术应用于检测系统的方法。实验表明,多线程技术提高了系统检测的效率。对于常见的表面贴装元件缺陷,在系统中分别采用了基于形状的匹配和基于组件的匹配进行检测。 最后给出了AOI系统的人机交互界面。本系统分为四个模块:图像采集模块、运动控制模块、检测算法模块和数据库管理模块,重点介绍了运动控制模块,并对各模块参数控制界面进行了设计。 With the development of electronic product moving towards high density and high performance, it brings forward more desire to facial quality inspection of electronic product. The image recognition-based automatic optical inspection /(AOI/), which is able to adapt to a high density mounted PCB, can detect defects quickly and reliably. The whole dissertation focuses on the specific research on facial quality inspection of printed circuit board assembly based on image process software HALCON8.0 and program platform VC++. The working principle of automatic optical inspection system is introduced simply. Then the hardware of AOI system is determinated and the specific components are picked out, which mainly includes light source, industrial camera, lens, servo motor and PCI-9014 control card. Image process software HALCON8.0 and program platform VC++ used in the system are introduced simply. Due to the restriction of the camera’s vision range, the method of image mosaic based on Harris is used to solve the problem which is that only the image of PCB’s local area can be captured at one time. An improved method of image mosaic is proposed based on the research of the three basic steps of image mosaic which is the inspection of feature points, matching of feature points and image fusion. Experimental results show this improvement method can shorten the time of the image mosaic, and this image can be used as a navigation picture of the inspection system. When the position of PCB would have offset and rotation relative with the standard one, the algorithm of affine transformation is applied to rectify image position to make the images alignment. The method of multithread technology applying in the inspection system is proposed based on the research of image match algorithm and multithread technology, which means image mosaic algorithm and image match algorithm are placed in the different thread. The application of multithread is able to improve the efficiency of inspection system and shorten the inspection time which has been testified by experiment. The method of shape-based matching and component-based matching are used in the inspection system to inspect the common facial defects of electronic product. Software interface of the AOI system is presented, which is divided into image acquisition module, motion control module, inspection algorithm module and database management module. All four modules are discussed in sequence specifically and the motion control module is also introduced in detail.

关 键 词： 自动光学检测 图像匹配 图像拼接 图像对准

分 类 号： [TP391.41]

领　　域： [自动化与计算机技术] [自动化与计算机技术]