飞机油箱内窥形貌建模与特征分析

发布时间：2018-03-27 05:14

本文选题：飞机油箱　切入点：图像采集　出处：《中国民航大学》2017年硕士论文

【摘要】：传统的飞机油箱检查方式是人工检查方式,工作人员往往需要钻入环境恶劣的油箱内部作业,危险性高且效率低下。为实现飞机油箱内部结构状态的视觉检测,课题对基于图像的飞机油箱内窥形貌建模方法及基于特征分析的缺陷检测方法进行研究,对提高飞机油箱检查效率,保证工作人员人身安全具有重要意义。第一,研究了飞机油箱内窥形貌全景图建模和缺陷检测过程中的图像获取问题。根据飞机油箱内部的结构特点及飞机油箱检修的需求,设计了飞机油箱内窥形貌观测系统,来完成图像采集;针对飞机油箱内窥形貌观测系统存在几何变形的问题,进行标定及畸变校正;研究了基于图像的飞机油箱内窥形貌全景图模型的三种投影方法。第二,设计了飞机油箱内窥形貌全景图建模过程中的图像拼接方案。拼接过程首先要对飞机油箱内窥图像进行特征点检测、初匹配和筛选;针对飞机油箱内窥图像的配准过程实质上属于不同平面内的场景匹配问题,研究了一种基于结构相似度的飞机油箱内窥图像配准算法;并研究了图像拼接效果的评价指标。第三,设计了基于特征分析的飞机油箱内窥图像缺陷判别方法。利用基于缺陷特征向量的判别方法判断飞机油箱图像中有无缺陷,而后根据复合特征分析方法找出缺陷分类依据并判断缺陷类别,最后采用基于运动恢复结构的单目定位算法对飞机油箱内部目标区域进行定位,为工作人员维修油箱提供参考。最后,对飞机油箱内窥形貌观测系统软、硬件平台进行了简要说明,并分别对本文所研究方法进行实验分析,验证了飞机油箱内窥形貌建模方法的有效性与缺陷识别方法的正确性。
[Abstract]:The traditional inspection method of aircraft fuel tank is manual inspection. The workers often need to drill into the interior of the fuel tank, which is dangerous and inefficient. In order to realize the visual inspection of the internal structure of the aircraft fuel tank, In this paper, the image based image modeling method of aircraft fuel tank and the defect detection method based on feature analysis are studied. It is of great significance to improve the efficiency of aircraft fuel tank inspection and ensure the personal safety of the staff. First, In this paper, the problem of image acquisition in the process of image modeling and defect detection of aircraft fuel tank topography is studied. According to the structural characteristics of aircraft fuel tank and the requirements of aircraft fuel tank inspection and repair, an aircraft fuel tank morphology observation system is designed. In order to achieve image acquisition, calibration and distortion correction are carried out in view of the problem of geometric deformation in the observation system of aircraft fuel tank morphology. Three projection methods for the panoramic image model of aircraft fuel tank endoscope topography based on image are studied. Second, The image splicing scheme in the modeling process of aircraft fuel tank topography panorama is designed. Firstly, the feature point detection, initial matching and screening of the aircraft fuel tank endoscope image are carried out. In view of the fact that the registration process of aircraft fuel tank endoscope image belongs to the scene matching problem in different planes, a new image registration algorithm for aircraft fuel tank endoscope image based on structural similarity is studied in this paper. The evaluation index of image stitching effect is also studied. Thirdly, an image defect discrimination method based on feature analysis is designed to judge whether there is any defect in the image of aircraft fuel tank by using the discriminant method based on defect feature vector. Then, according to the composite feature analysis method, the defect classification basis is found and the defect category is judged. Finally, a monocular localization algorithm based on motion recovery structure is used to locate the target area inside the aircraft fuel tank. Finally, the software and hardware platform of the observation system for the morphology of the aircraft fuel tank is briefly described, and the experimental analysis of the methods studied in this paper is carried out. The validity of the modeling method and the correctness of the defect recognition method are verified.
【学位授予单位】：中国民航大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：V228.11;TP391.41

【参考文献】