基于机器视觉的激光打标工件定位系统研究

发布时间：2018-04-05 03:19

本文选题：激光打标　切入点：机器视觉　出处：《华中科技大学》2016年硕士论文

【摘要】：激光打标是激光加工领域中应用最为广泛的,打标过程中工件位置的确定是关键的一步,随着生产的发展,使用固定夹具对工件进行定位的方式不仅成本较高,且不利于多工件同时定位,同时也会带来人为定位误差的不稳定性。为此提出了将视觉定位技术引入到激光打标系统中,利用视觉定位技术的优点实现工件快速稳定准确的定位,提高生产效率。本文首先对激光打标和机器视觉技术的发展做了简单介绍,调研分析了目前国内外的发展状况。其次结合工件图像的具体特征研究分析了本文用到的图像处理相关算法,如图像分割、二值形态学图像滤波、二值图像边缘检测等。最后研究了视觉定位中的相机标定技术,考虑到传统标定算法的复杂度高,操作过程繁琐,以及激光打标过程中工件的定位是在一个二维平面内进行的,因此提出使用一种基于单平面的射影变换的相机标定方法,并引入多项式矫正算法对成像过程中产生的畸变进行矫正。为了获得工件在激光打标坐标系中的位置和方位角,本文在前期图像处理的基础上提取工件轮廓,利用轮廓的中心来确定工件位置,结合最小二乘法对提取到的轮廓使用Hough变换精确地检测出轮廓中的直线,并根据直线的倾角得到工件的方位角。根据实际的应用需求,分析总结系统各个硬件部分的选取原则,合理搭建定位系统的硬件部分。在VS2005下利用微软的基础类库(MFC)进行本系统软件界面的设计,并使用多线程技术提高软件界面的响应速度,同时使用德国SCAPS公司提供的Activex控件使定位软件与打标系统的控制软件SamLight进行数据交换,最终实现打标系统的工件定位。在整个系统的软硬件搭建完成后进行实验,相机标定实验结果表明针对本系统的标定算法能够取得较高的标定精度,x、y方向上的平均标定误差在0.1mm以内,最大标定误差均小于0.1mm。对标刻平台上的工件进行定位打标,实验表明本定位系统能够在系统所允许的误差内对工件进行准确稳定的定位。最后,本文从图像处理精度方面深入分析了引起定位误差的原因,为本文将来的工作奠定基础。
[Abstract]:Laser marking is the most widely used in the field of laser processing. The determination of the position of the workpiece in the marking process is a key step. With the development of production, it is not only costly to use fixed fixture to locate the workpiece.It is not conducive to simultaneous localization of many jobs, but also to the instability of artificial positioning errors.In this paper, the visual positioning technology is introduced into the laser marking system, the advantages of the visual positioning technology are used to realize the rapid, stable and accurate positioning of the workpiece, and the production efficiency is improved.In this paper, the development of laser marking and machine vision technology is briefly introduced, and the development situation at home and abroad is analyzed.Secondly, the image processing algorithms used in this paper, such as image segmentation, binary morphological image filtering, binary image edge detection and so on, are analyzed in combination with the specific characteristics of the workpiece image.Finally, the camera calibration technology in visual positioning is studied. Considering the complexity of the traditional calibration algorithm, the complicated operation process, and the localization of the workpiece in the laser marking process is carried out in a two-dimensional plane.Therefore, a camera calibration method based on single plane projective transformation is proposed, and polynomial correction algorithm is introduced to correct the distortion in the imaging process.In order to obtain the position and azimuth angle of the workpiece in the laser marking coordinate system, the contour of the workpiece is extracted on the basis of the previous image processing, and the position of the workpiece is determined by the center of the contour.The straight line in the contour is accurately detected by using Hough transform and the azimuth of the workpiece is obtained according to the dip angle of the straight line by using the least square method.According to the actual application demand, the selection principle of each hardware part of the system is analyzed and summarized, and the hardware part of the positioning system is set up reasonably.The software interface of this system is designed by using Microsoft basic Class Library (MFC) under VS2005, and multithreading technology is used to improve the response speed of the software interface.At the same time, using the Activex control provided by SCAPS company in Germany, the positioning software can exchange data with the control software SamLight of marking system, and finally the work piece location of marking system can be realized.After the hardware and software of the whole system are finished, the camera calibration experiment results show that the calibration algorithm can achieve high calibration accuracy in the direction of 0.1mm, and the maximum calibration error is less than 0.1 mm.The experiments show that the positioning system can locate the workpiece accurately and stably within the allowable error of the system.Finally, this paper analyzes the cause of the location error from the aspect of image processing accuracy, and lays a foundation for the future work of this paper.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.41;TN249

【相似文献】