【作者】 李君兰；

【导师】 张大卫；

【作者基本信息】 天津大学 ， 机械制造及其自动化， 2010， 博士

【摘要】 随着引线键合技术向着细密化、高精度的方向发展,视觉系统已经成为影响IC封装精度的重要因素之一。本论文以面向芯片封装的视觉系统作为研究对象,根据芯片封装设备对视觉系统的要求搭建视觉系统架构,并对系统标定方法、视觉定位算法以及视觉系统与控制系统的集成等进行了全面的研究,取得的创造性成果如下:1)针对芯片封装对视觉系统成像质量、放大倍率和工作物距要求,对光学系统进行了设计,其具有两级放大结构和特殊的照明方式。分别采用两种镜头对视觉系统进行了构建,并通过实验验证了视觉系统的成像效果,实验表明两套视觉系统均获得满意的拍摄效果。2)提出了一种用于平行成像系统的基于几何矩的近平行系统摄像机标定方法。该算法解决了平行成像系统的摄像机标定问题,使其通过获取单幅模板图像即可有效地求解全部摄像机参数。仿真和实验结果表明标定后得到的摄像机参数结果准确稳定,且精度较高。3)提出了使用带有网格的分划板对显微视觉系统进行标定的方法。实验表明,针对显微视觉系统的标定方法稳定有效,能够很好地适用于此前难于解决的显微视觉系统的标定问题。4)针对芯片封装时使用的芯片及其图像的特点,分别提出了适合芯片和引线框架的定位算法。对于芯片焊点,采用快速傅里叶变换算法进行匹配定位;对于芯片的引线框架,采用基于轮廓提取的不变矩方法进行匹配定位。经过实验验证,这两种方法的定位精度能够满足芯片封装的要求。在Labview中对匹配定位进行了编程,并实现了在Labview中对芯片和引线框架的视觉定位。5)给出了视觉坐标系与运动坐标系间的坐标变换关系、误差补偿模型以及视觉反馈框图。搭建了视觉与运动控制集成后的实验系统,进行了系统标定,并通过实验建立了误差补偿模型。在此基础上,视觉系统即可与控制系统进行准确的坐标转换,从而通过视觉定位算法获得的芯片位置误差从视觉坐标转换到运动坐标系,实现精确的定位。更多还原

【Abstract】 With the requirements of high accuracy and high speed in wire bonding, machine vision positioning system has become one of important factors which influence the quality of IC packaging. In this case, it can be used as a feedback of control system to achieve accurate positioning of chips. A positioning system using machine vision technologies has been designed and implemented in this dissertation to obtain accurate positioning. The calibration method and visual localization algorithm are researched, and the integration of vision and motion is studied. The main works and contributions in this dissertation are as follow:1) In terms of hardware of vision system, the structure of the optical system has been designed based on analyzing packaging equipment’s requirements. The hardwares of vision system are selected base on two different optical lenses. In order to verify the imaging effect, experiments are done, which indicate that clear images can be obtained using vision systems2) In terms of camera calibration method, this paper presents a novel approach of camera calibration under the near parallel condition based on geometric moments. The camera calibration method is carried out by using a calibration board with arranged rectangular features. Before calibration, feature regions on the image of calibration board are extracted, and centroid and rotation of each feature are detected using geometric moments. Then, the closed-form solutions to the parameters are provided by combining Gauss lens model with camera model, and all camera parameters are refined by a nonlinear minimization procedure according to the distortion coefficients. Simulations and experiments are performed to verify the proposed camera calibration algorithm. The experimental results indicate that the algorithm is feasible with a printed calibration board and simple operating steps, and the the precision of camera calibration is high.3) A calibration method for micro-vision system is present using a reticule with grid as calibration pattern. Experiments are performed which show that the calibration of micro-vision system is reliable. Beacause of the calibration pattern is convenient to obtain, this method has practicability.4) In terms of visual localization algorithm, the method of precision positioning for chip and its leadframe is present. The normalized Fast Fourier Transform is applied for locating weld spots on chip, and a positioning method with contour extraction based on invariant moments is proposed for locating leadframe. According to experimental results, the precision of these methods can satisfy packaging requirements. The localization is implemented by Labview program.5) In terms of system integration, the experiment system is established by combining vision and motion system. Coordinate transformation between vision and motion coordinate system, error compensation model and block diagram of visual feedback are proposed. On the basis, the position error, calculated by visual localization algorithm, is converted from vision coordinate to motion, and high precision positioning is achieved.更多还原