基于图像处理的转辙机接点组打入深度提取算法
Penetrating Depth of Contact Group Extraction Algorithm for Switch Machine Based on Image Processing
DOI: 10.12677/CSA.2021.1112292, PDF,    国家自然科学基金支持
作者: 杨晨威, 徐尚志, 李志鹏：同济大学信息与通信工程系，上海
关键词: 转辙机；图像测量；图像滤波；颜色分割；Switch Machine； Image Measurement； Image Filtering； Color Segmentation
摘要: 转辙机是铁路上改变轨道方向，反映道岔位置的重要器械，转辙机腔内接点组中动接点与静接点的打入深度是判断转辙机是否正常工作的重要指标。由于转辙机内部结构非常复杂，传统工业物件的图像测量方法并不适用，为了提高转辙机检修的准确性、减小人力开销，本文设计了基于图像处理技术，针对广泛使用的ZD(J)9型转辙机的内部打入深度的测量算法。首先该算法通过模板匹配的方式准确定位整个接点组区域，排除其他区域结构的干扰。同时针对拍摄图像可能存在的倾斜问题，提出了基于图像仿射变换的预处理算法将倾斜的接点组图像校正。从图像距离到实际距离需要一个比例尺进行换算，算法在校正后的图像中选取长度明确、畸变较小的物体长度作为比例尺计算的参照物。为进一步精确定位动接点和静接点的位置，设计了基于meanshift的滤波算法，HSV色彩空间中的颜色阈值分割算法，以及按照数字图像连通域的去噪算法。最后，在精确定位的基础上加以一定误差补偿，进行参数计算。本文算法在不同拍摄高度下得到的图像中进行实验，对比了不同拍摄高度下的参数提取效果，对误差原因进行详细分析。
Abstract: Switch machine is an important instrument to change track direction and reflect switch position in railway. The penetrating depth between dynamic contact group and static contact group is an important index to judge whether switch machine worked normally. Owing to complicated internal structure of switch machine, traditional image measurement methods of industrial objects are not applicable. In order to improve the accuracy of switch machine maintenance and reduce the cost of manpower, this paper designed an algorithm for measuring the internal penetrating depth of ZD(J)9 switch machine based on image processing. First of all, this algorithm can accurately locate the whole contact group region by means of template matching, which eliminated the interference of other region structures. Meanwhile, a preprocessing algorithm based on affine transformation is also proposed to rectify oblique images, after which the object length with clear length and small distortion will be selected as the reference for scale calculation used for converting image distance to actual distance. Some methods help further accurately locate the positions of contact groups including meanshift filtering, color threshold segmentation in HSV color space and a denoising algorithm based on connected domain were also proposed. Finally, with the help of error compensation algorithm, the calculation result can be more precise. This algorithm was tested on images obtained at different shooting heights, revealing the effect of parameter extraction by comparison. The causes of error were also analyzed in detail.
文章引用：杨晨威, 徐尚志, 李志鹏. 基于图像处理的转辙机接点组打入深度提取算法[J]. 计算机科学与应用, 2021, 11(12): 2872-2889. https://doi.org/10.12677/CSA.2021.1112292

参考文献

[1] Khan, S. (2001) Object Based Segmentation of Video Using Color and Spatial Information. IEEE Computer Vision and Pattern Recognition Conference, Kauai, 8-14 December 2001, 746-751.
[2] Elias, N. (2003) A Survey on Industrial Vision Systems Applications and Tools. Image and Vision Computing, 21, 171-188. [Google Scholar] [CrossRef] 
[3] Zhang, Y. (2004) Deformation Visual Inspection of Indus-trial Parts with Image Sequence. Machine Vision and Application, 15, 115-120. [Google Scholar] [CrossRef] 
[4] 汤勃, 孔建益, 伍世虔. 机器视觉表面缺陷检测综述[J]. 中国图象图形学报, 2017, 22(12): 1640-1663.
[5] 孙国栋, 赵大兴. 机器视觉检测理论与算法[M]. 北京: 科学出版社, 2015.
[6] 张云辉, 谭庆昌, 田原嫄. 图像测量系统精度影响因素的研究[J]. 微计算机信息, 2008, 8(2): 271-273.
[7] 吴晓波, 等. 图像测量系统中的误差分析及提高测量精度的途径[J]. 光学精密工程, 1997, 5(1): 133-141.
[8] 陈向伟, 王龙山, 刘庆民. 图像测量技术在微型齿轮测量中的应用[J]. 半导体光电, 2004, 25(6): 465-468.
[9] 王心悦. 基于机器视觉的轴承圆柱滚子尺寸测量关键技术研究[D]: [硕士学位论文]. 哈尔滨: 哈尔滨理工大学, 2021.
[10] Gadelmawla (2011) Computer Vision Algorithms for Measurement and Inspection of Spur Gear. Measurement, 44, 1669-1678. [Google Scholar] [CrossRef] 
[11] He, J. (2010) Size Detection of firebricks based on machine vision technology. International Conference on Measuring Technology and Mechatronic Automation, Changsha, 13-14 March 2010, 2648-2651.
[12] 潘强. 车辆三维尺寸的图像测量方法研究[D]: [硕士学位论文]. 西安: 长安大学, 2019.
[13] Li, S., Jiang, X. and Qian, H. (2016) Vehicle 3-Dimension Meas-urement by Monocular Camera Based on License Plate. IEEE International Conference on Robotics and Biomimetics, Qingdao, 3-7 December 2016, 800-806. [Google Scholar] [CrossRef] 
[14] 梁玉亮. 基于图像的树木三维建模与树木参数提取[D]: [硕士学位论文]. 哈尔滨: 东北林业大学, 2019.
[15] 管焱然, 管有庆. 基于OpenCV的仿射变换研究与应用[J]. 计算机技术与发展, 2016, 26(12): 58-63.
[16] Canny, J. (1986) A Computational Approach to Edge Detection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 8, 679-698. [Google Scholar] [CrossRef] 
[17] He, K.M. and Tang, X.O. (2013) Guided Image Filtering. IEEE Transactions on Pattern Analysis and Machine Intelligence, 35, 1397-1409. [Google Scholar] [CrossRef] 
[18] Camaniciu (2002) Mean Shift—A Robust Approach toward Feature Space Analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence, 24, 603-619. [Google Scholar] [CrossRef] 
[19] Sural, S. (2002) Segmentation and Histogram Generation Using the HSV Color Space for Image Retrieval. International Conference on Image Processing, Rochester, 22-25 September 2002, 589-592.
[20] Otsu, N. (2007) A Threshold Selection Method from Gray-Level Histograms. IEEE Transaction on Pat-tern Analysis and Machine Intelligence, 9, 62-66. [Google Scholar] [CrossRef] 
[21] Duda, R.O. and Hart, P.E. (1972) Use of the Hough Transformation to Detect Lines and Curves in Pictures. Communications of the ACM, 15, 11-15. [Google Scholar] [CrossRef] 
[22] Bai, H.L. and Liu, C.P. (2004) A Hybrid License Plate Extraction Method Based on Edge Statistic and Morphology. International Conference on Pattern Recognition, Cambridge, 23-26 August 2004, 831-835.