【作者】 张稳；

【导师】 杨光；

【作者基本信息】 北京邮电大学 ， 机械制造及其自动化， 2013， 硕士

【摘要】 自动检测系统的设计与实现在实际生产中具有重要的意义。随着计算机视觉技术的快速发展,工业自动化检测已被列为世界各国着重的发展对象。目前疵病检测技术在工业自动化检测中是一个新兴的方向,它综合不同领域的背景知识,检测技术和检测方法都在进一步的探索和试验中。本文研究了一个针对于枪械内膛的疵病检测系统。枪械内膛的制造工艺,使得枪械内膛的疵病检测具有较为复杂的背景环境。算法的设计与实现主要是针对于这种复杂的检测环境。算法的执行过程中主要用到了彩色图像的灰度化、二值化、连通区域的标记、小面积疵病的统计和去除、连通区域的特征参数的提取、判决识别较大面积的疵病、图像细化和角点检测等相关技术。算法分阶段分步骤的将图像中的不同疵病识别或判决出来。连通区域特征参数的提取主要用到了连通区域的矩形度、圆形度、不变矩和长宽比等变量,通过这些变量不同的阈值将疵病分离出来。在检测系统的设计与实现中,还涉及到图像的采集和存储,串口编程,系统界面设计等相关的编程方法。疵病检测算法能够检测到疵病的严重程度,从定性和定量的角度给出检测的结果值。能够识别出80%以上的疵病,具有实际应用价值。更多还原

【Abstract】 Design and implementation of automated detection system has significance in actual production. As the rapid development of computer vision technology, the industrial automation detection has been listed as the object of development which the world focuses on. Defect detection is an emerging technology in the direction of industrial automation detection, it is combined with different professional background in the field of knowledge, detection technology and method is in further explored and experiment.In this paper, defect detection system of the barrell is studied. Due to the manufacturing process, the defect detection system has a more complex content. In order to achieve a better non-destructive testing, a detection algorithm based on image processing for barrel wall is designed in the paper. The design of algorithm mainly use image graying, image binarization, marking connected region, defect of little area statistics and remove, image thinning, corner detection technology and so on. This algorithm process is designed to solve different defect types. The extraction of characteristic parameter is mainly used in the connected region squareness, circularity, the invariant moment, aspect ratio and other variables, this method can be separated from the defect and rifling. In the implementation and design of detect system, image acquisition and storage will be involved, and the same with system interface design and serial programming.The various functions of the system already completely implement, the defect detection algorithm is able to detect the severity of the defect, and reference value is given from the view of qualitative and quantitative point. This algorithm can recognize more than eighty percent defect, and have actual use value.更多还原