飞机蒙皮缺陷机器视觉检测技术研究

发布时间：2018-11-16 16:21

【摘要】：随着经济和技术的发展,机器视觉技术已经代替人眼深入到了社会的方方面面,彻底改变了人们的生活环境。机器视觉检测综合了机器视觉和自动化技术,广泛应用于制造行业的产品缺陷检测,例如产品的装配过程检测与定位、产品的包装检测、产品的外观质量检测、物流行业的货物分拣或水果分拣等,机器视觉能够代替人工快速、准确地完成各项工作。本文针对老龄飞机蒙皮缺陷不易检测问题,分析现有检测技术的优缺点,采用机器视觉、DSP、模式识别等技术,在机器人平台上完成了老龄飞机蒙皮缺陷检测系统的构造,本系统通过无线传输技术将采集的实时、动态蒙皮损伤信息传递给地面健康监测平台进行分析,可在线得到缺陷检测结果,实现了飞机蒙皮缺陷图像分类和铆钉连接关键部位腐蚀图像分类。老龄飞机蒙皮缺陷监控系统,主要包括图像采集、无线通讯、图像存储、图像处理、特征提取、模式识别六个模块。根据系统的要求,完成了图像采集模块、无线通讯模块、图像存储的硬件设计和图像处理模块、特征提取模块、模式识别模块的软件设计。针对飞机蒙皮缺陷特征提取,建立了飞机蒙皮图像样本库,采用了灰度矩阵法提取飞机蒙皮缺陷图像特征值,提取的特征值精度满足系统要求；针对铆钉连接部位腐蚀图像特征提取,给出了一种改进的铆钉中心定位算法确定腐蚀铆钉的中心与半径,改进的算法提高了铆钉中心确定的准确度,进而提高了铆钉连接部位腐蚀图像特征值的精确度。在模式识别模块,阐述了一般线性支持向量机、一般非线性支持向量机和模糊支持向量机的原理与分类应用,本系统选择了模糊支持向量机进行模式识别,在基于样本中心距离FSVM方法的基础上,给出了基于样本间距FSVM方法进行飞机蒙皮图像和铆钉连接部位腐蚀等级进行分类,通过仿真实验对比发现该算法-定程度地提高飞机蒙皮缺陷图像和铆钉腐蚀蒙皮图像的识别分类率。基于机器视觉的飞机蒙皮缺陷监控系统与目前的检测设备相比,无需专业操作人员就可在计算机前完成检测,具有良好的扩展性,应用前景广阔,对提高老龄飞机的可靠性有较大的应用意义。
[Abstract]:With the development of economy and technology, machine vision technology has taken the place of human eyes to go deep into all aspects of society and completely changed people's living environment. Machine vision inspection, which integrates machine vision and automation technology, is widely used in product defect detection in manufacturing industry, such as product assembly process detection and positioning, product packaging testing, product appearance quality testing, Goods sorting or fruit sorting in the logistics industry, machine vision can replace manual fast, accurate completion of the work. Aiming at the problem that it is difficult to detect the skin defects of the aged aircraft, this paper analyzes the advantages and disadvantages of the existing detection techniques, and uses machine vision, DSP, pattern recognition and other technologies to complete the construction of the skin defect detection system of the aged aircraft on the robot platform. Through wireless transmission technology, the real-time and dynamic skinning damage information is transmitted to the ground health monitoring platform for analysis, and the defect detection results can be obtained online. Aircraft skin defect image classification and rivet joint key part corrosion image classification are realized. The skin defect monitoring system of aging aircraft mainly includes six modules: image acquisition, wireless communication, image storage, image processing, feature extraction and pattern recognition. According to the requirements of the system, the software design of image acquisition module, wireless communication module, image storage hardware and image processing module, feature extraction module, pattern recognition module is completed. Aiming at the feature extraction of aircraft skin defect, the sample database of aircraft skin image is established, and the gray matrix method is used to extract the feature value of aircraft skin defect image, and the accuracy of the extracted feature value meets the requirements of the system. An improved rivet center location algorithm is presented to determine the center and radius of rivets. The improved algorithm improves the accuracy of rivet center determination. Furthermore, the accuracy of the characteristic value of the corrosion image of rivet joint is improved. In the pattern recognition module, the principle and classification application of general linear support vector machine, general nonlinear support vector machine and fuzzy support vector machine are expounded. Based on the FSVM method of sample center distance, the aircraft skin image and the corrosion grade of rivet connection are classified based on sample spacing FSVM method. The simulation results show that the algorithm can improve the recognition rate of aircraft skin defect image and riveted skin image to a certain extent. The aircraft skin defect monitoring system based on machine vision can complete the inspection in front of the computer without professional operators, compared with the current testing equipment. It has good expansibility and broad application prospect. It has great application significance to improve the reliability of aging aircraft.
【学位授予单位】：长春工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.41

【参考文献】