铁路隧道限界检测中断面图像处理技术研究

发布时间：2018-05-21 15:08

本文选题：隧道限界检测 + 图像处理　；参考：《北京交通大学》2017年硕士论文

【摘要】：隧道限界检测对保障铁路安全具有重要意义。随着近年来高速铁路的大量建设,基于图像处理的自动化检测方法逐渐成为当前主流技术。电视摄像测量法由于具有操作简便、测量精度高等优点,成为诸多测量方法中应用最为广泛的技术,其中,从断面图像中快速准确地提取激光线是整个过程尤为关键的一步,将直接关系到限界计算的准确性。然而,隧道中存在的不均匀光照、线缆、内壁纹理等复杂背景为激光线轮廓的提取带来了诸多技术难题。本文利用图像处理相关技术,研究复杂背景下断面激光线轮廓的提取与后处理算法。本文充分考虑复杂隧道背景对限界测量的影响,提出一种适用于多种场景的激光线轮廓提取算法。采用改进的局部阈值分割技术对断面图像进行预处理,并结合边缘提取与直线检测算法检测图像中线型特征物,再通过像素灰度值分析与眼动跟踪实验相结合方法提取激光线轮廓。大量隧道检测结果表明,本文提出的轮廓提取算法有较强的稳定性和较高的准确率,为后续限界测量与超限判断奠定了良好基础。针对激光线提取过程中存在的宽度过粗、曲线断开等现象,提出一种激光线后处理方案。选取适用于本场景的细化算法处理激光线图像得到单像素轮廓,再提出两种平移方案对单像素骨架线的位置做出修正,分别给出两种方案的方差表达式及仿真实验结果,选取误差较小方案对骨架线位置进行平移修正;最后提出一种基于连通区域基本段的拟合生长算法,对经细化与平移处理的激光线轮廓断开部分进行拟合。实验结果表明,该方案能在误差允许范围内实现对激光线断开、缺失处的拟合生长,符合现代铁路隧道的检测要求。
[Abstract]:With the construction of high speed railway in recent years, the automatic detection method based on image processing has gradually become the mainstream technology with the construction of high speed railway in recent years. The video camera measurement method has become the most widely used technology in many measuring methods because of its advantages of simple operation and high measurement precision. The fast and accurate extraction of laser line from the cross section image is a crucial step in the whole process, which will directly affect the accuracy of the limit calculation. However, the complex background of the tunnel, such as uneven illumination, cable and inner wall texture, is a difficult problem for the extraction of laser line contour. This paper uses image processing techniques. In this paper, an algorithm for extracting and post-processing the contour of a laser line in a complex background is studied. This paper gives a full consideration of the influence of the complex tunnel background on the limit measurement, and proposes a laser line contour extraction algorithm suitable for various scenes. The improved local threshold segmentation technique is used to preprocess the image of the cross section, and the edge extraction and the straight line are extracted. The detection algorithm detects the linear features in the image, and then extracts the contour of the laser line by combining the pixel gray value analysis with the eye movement tracking experiment. A large number of tunnel detection results show that the proposed contour extraction algorithm has a strong stability and high accuracy, which lays a good foundation for the follow-up limit measurement and the overlimit judgment. A laser line post processing scheme is proposed for the phenomenon that the width of the laser line is too rough and the curve is disconnected. A thinning algorithm suitable for this scene is selected to process the single pixel contour of the laser line image. Then two translation schemes are proposed to modify the location of the single pixel skeleton line, and the variance expression of the two schemes is given respectively. The result of the model and the simulation experiment, select the small error scheme to make the shift correction to the position of the skeleton line. Finally, a fitting growth algorithm based on the basic segment of the connected region is put forward to fit the disconnected part of the laser line by the thinning and translation. The experimental results show that the scheme can realize the laser line breaking in the error allowable range. The fitting growth at the missing points accords with the requirements of modern railway tunnels.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U459.1;TP391.41

【参考文献】