【作者】 余琨；

【导师】 严国萍；

【作者基本信息】 华中科技大学 ， 电路与系统， 2007， 硕士

【摘要】 机器视觉是近年来发展迅速的一门综合性学科。计算机硬件和软件技术的飞速发展,尤其是高性能、高分辨率数字影像输入设备的出现,为机器视觉技术在工业自动化、工业在线检测、精密测量等领域的应用提供了良好的软硬件基础。本课题是为中国电光研究所开发研制的基于机器视觉的双目头盔显示器检查仪系统。双目头盔显示器是一种新型的机载显示和瞄准系统,双目视场采用100%全叠加视场。通过分别检测左目图像和右目图像中的圆型和十字形目标的位置,来判断双目视场是否调整合格。本文从需要解决的问题和研究现状出发,根据机器视觉的原理和设计方法确定了检查仪设计方案。该系统中采用CMOS摄像头作为图像获取装置,以获取头盔显示器左目和右目的图像;利用计算机串口,通过电动运动控制器远程控制图像采集装置的移动。首先利用CMOS摄像头采集大平行光管的图像,作为标准图像。检测标准图像中的目标位置作为标准位置。然后利用CMOS摄像头采集头盔显示器左目和右目的图像,利用数字图像处理技术对采集的图像中圆形和十字形目标进行定位。将得到定位结果和标准图像中目标的位置进行比较分析,得到头盔显示器双目图像中目标的偏差,根据偏差数据,就可以判断头盔显示器是否调整合格。系统中可将数据生成文本格式的报表,作为产品检查记录。本系统软件开发采用MFC框架结构,提高了软件开发的灵活性。串口控制是利用MsComm控件实现,极大地提高了运动控制器远程操作的方便性。为了实现图像中目标定位,文中分析比较了几种图像处理的方法,通过实验比较分析,确定了目标定位的图像处理方法。首先对目标图像进行了灰度增强处理,以提高图像对比度;然后通过图像阈值分割,将目标与背景分离;最后根据目标特点,采用相关法对目标进行识别并利用模版匹配法和灰度重心法对目标进行定位。该项目已实际制作调试完毕,验收合格,交付用户使用。更多还原

【Abstract】 Machine vision is a subject which has developed rapidly in recent years. The development of computer hardware and software technology, especially the appearance of high performance and high distinguish rate vision input device, have provided solid base for the machine vision’s usage in the fields of industrial automation, industrial online test, precision measurement and etc.The test system of Helmet-Mounted Displays is an application which based on machine vision. Helmet-Mounted Displays is a new system for aim on airplanes. The pictures of left-side display and right-side display are taken independently. And the round and crisscross object’locations in the pictures were acquired. Finally, whether the Helmet-Mounted Displays were adjusted eligible is judged by the results.The concept, characteristic and structure of machine vision were briefly introduced in this article. The scheme of the test system was also researched, including the hardware structure and software function. The CMOS was used in this system as image-acquire device and controlled through serial-port. The pictures of left-side display and right-side display are taken by CMOS. And the objects’locations in the pictures were detected by digital image processing technology. The report of the results was made in text format.The software of the system was developed in MFC frame. The controlling of serial port was realized by MsComm. Through many experiments, grayscale transformation, threshold splitting and template matching was used to find the locations of the objects in the image processing.This project has been completed eligibly and consigned to consumer.更多还原