基于图像处理的125MN挤压机对中检测系统设计与研究

【作者】 吴士旭；

【导师】 谭建平；

【作者基本信息】 中南大学 ， 机械电子工程， 2008， 硕士

【摘要】 基于数字图像处理技术的工业测量在国外已经得到了深入研究和广泛的应用,在国内也受到了越来越多的重视。随着计算机硬件性价比的不断提高,相关技术的不断发展,此项技术克服了传统测量技术精度差、效率低的不足,以其快速、非接触、精确等特点,成为工业现场检测的一种重要检测手段。本文以125MN挤压机改造项目作为研究背景,突破传统的微位移检测手段,采用数字图像处理技术作为检测方法,并研究挤压机对中检测系统中的图像的远程接入控制。特别针对USB接口摄像机在工业现场运用中存在的不足,设计摄像机控制系统以解决系统图像采集中的传输距离、摄像机选择和稳定性三大问题的关键技术。以Visual C++为开发平台,以Opencv(Intel Open Source Computer Vision Library)作为辅助开发工具,利用图像处理技术,根据挤压中心检测算法模型,实现挤压机的对中检测。在课题涉及的数字图像处理技术中,图像分割和摄像机标定是测量系统的关键技术环节。论文通过算法模型分析处理两路激光光束在成像装置中的位置来获取挤压机挤压中心,首要任务就是实现激光光斑与背景图像的分割,论文中通过比较各种图像分割算法模型,在图像预处理的基础上,采用大津法实现图像分割,然后应用重心法获得激光光斑位置坐标。在上述图像处理的过程中,如何消除摄像机的镜头加工及安装误差带来的图像畸变的方法成为提高检测精度的重要途径。通过比较各种标定技术理论方法,本文运用传统摄像机标定技术中的基于平面模板的摄像机标定技术,重点研究了摄像机内参中的畸变系数的求解方法,利用获得的摄像机畸变系数对图像进行校正。并分析了检测系统中硬件设备的安装过程中所带来的倾斜对检测结果带来的影响。以上述理论方法作为研究基础,苯论文从实际应用出发,将数字图像处理技术和微处理器控制技术结合起来,开发了挤压机对中检测系统中的图像采集和处理部分。系统应用实践表明,采用论文所设计的摄像机控制传输和图像处理理论方法,达到了对中检测系统的图像处理技术要求。更多还原

【Abstract】 The industry measurement based on digital image processing technology abroad has been studied and applied extensively, in country which has also been taken more and more attention. With the continuous decrease of computer hardware cost and gradual development of related technology, the technology overcomes the traditional measurement of low accuracy and inadequate efficiency. Its fast, non-contact, precision, and other characteristics become an important means of industrial detection method. In this paper ,it takes the 125 MN extruder reconstruction project as a research background, brokers through the traditional means of micro-displacement detection methods, and studies the camera remote access and control of extrusion detection system based on digital image detection technology. Special for the deficiencies of the USB interface camera in the industrial scene use, camera control system designed to solutions three key technologies: remote transmission of the image acquisition system, the camera selection and the system working stability. It takes Visual C + + as the development platform ,and Opencv (Intel Open Source Computer Vision Library) as a complementary development tools, using image processing technology, according to squeeze Centre detection algorithm model, to realize center detection of the extrusion machine.In this paper, involved in the digital image processing technology, image segmentation and calibration measurement system are the key technology sectors. Through algorithm and process model of two laser beams in the imaging device to obtain the position of Extruder extrusion Centre, the first and foremost task is to achieve the laser spot and the background image segmentation. In the thesis, by comparing the various image segmentation algorithm models, using the Otus segmentation method to achieve the image segmentation based on the pre-image processing, and then the center-gravity-method was applied to get the laser spot position coordinates. During above image processing, the method how to eliminate the manufacturing and installation error of camera lens distortion of the image becomes an important way of improving the accuracy of detecting. Through comparing various camera calibration theory methods, the paper uses the traditional paper camera calibration technique based on the plane of the camera calibration template technology, focuses on the camera internal parameters of the solution aberration coefficient, and then the distortion ratio was used to correct the camera distortion coefficient Image. The system in the installation of hardware equipment in the course of the tilt of the impact test results was analyzed. To the theoretical basis for research topics from a practical project application, by using the digital image processing and microprocessor control technologies, developed the detection system in the collection and processing of the images for the center detection of extrusion machine. System practice application shows that the design of camera control and transmission methods meets the requirements of center detection system with the image processing theory technology.更多还原