【作者】 姚峰林；

【导师】 李元宗；

【作者基本信息】 太原理工大学 ， 机械电子工程， 2005， 硕士

【摘要】 机器视觉是近年发展起来的一个新兴学科,以它为基础的CCD应用技术近年来由于其广泛应用而受到学术界的青睐。基于CCD的机器视觉系统以组成简单和使用方便,近年来在非接触测量系统中应用更为广泛。作为机器人技术不可或缺的部分机器视觉和CCD也就成了我们研究的重点。 本文以铁轨对中性检测为目的,从介绍机器视觉相关的光学与镜头知识为课题的开端,着重介绍了机器视觉中的光源,照明以及镜头问题。作为近年来机器视觉中的成像的最主要的器件,CCD近年来也有了突飞猛进的发展,本课题的主要实验器件也是CCD,所以它的工作原理和器件特性也成为本课题研究的重点之一。基于我们实验室现有的设备,我们做了基于灰度的图像匹配,基于轮廓的图像匹配,以及基于聚类法的彩色图像分割法在铁轨对中性检测中的应用。并对它们在应用中的优劣进行了研究与分析,而且取得了一些具有一定意义的结论,为我们今后的工作指明了方向。 作为无接触测量的重要部分的标定技术以及测量,误差的探讨也成为本课题的重要研究问题。我们从理论出发结合实际应用,分析了测量系统中的误差来源和它们对测量的影响,并以齐次变换为基础,得到了图像空间与物空间的坐标变换方更多还原

【Abstract】 Machine vision is a newly developed subject in recent years. The CCD application technology based on machine vision is favored by the academe because of its abroad application. Machine vision system based on CCD is widely applied in non-contact measurement for its simple composing and convenient applying.As an indispensable part of robotics, machine vision and CCD become the emphasis of our study.In order to realize the track center inspect, the paper introduce the optics and lens which are used in machine vision as the beginning. It laid emphasis on the lamp house, illumination and lens used in machine vision. As the most commonly used appliance -CCD, CCD has rapidly developed in recent years. It is also the central experimental apparatus. So its working mechanism and trait also become a question to be discussed.By using available apparatus in our laboratory, we have done the image matching based on gray level, one on contour image matching, and multicolor image partition on clustering. We haveanalyze the advantage and disadvantage, and also made some valuable conclusion, and point out the way to solve the problem in future.As the important part of non-contact measurement, the calibration technology and error analysis are also important questions to study. Based on the theory and combined with the actual application, the paper analyze the source of error in the inspect system and the effect which affected. Based on the homogeneous transformation, we deduct the equation of transformation between image spaces coordinate and inspect space coordinate. It will direct the actual applied calibration.Applying machine vision and CCD in track center location inspect is the first time to bring out such concept and method. Although the study on the question need further study and perfect, it give us much more wide developing foreground.更多还原