基于机器视觉的无砟轨道层间结构位移测量方法研究

为实现板式无砟轨道层间离缝的全天候监测,提出一种基于机器视觉的非接触式层间结构位移测量方法。针对传统Canny边缘检测算法自适应性较差的问题,对其进行改进。利用自适应中值滤波器代替高斯滤波器;提出一种基于分块迭代的局部阈值分割法自适应地选取阈值,并通过高斯拟合算法检测改进算法的优越性。通过室内试验对标记点大小、光源强度、拍摄距离和拍摄角度4个方面进行对比分析,选择最适测量参数。试验结果表明,在标记点运动的状态下,最大测量误差为2.6%,在误差的允许范围之内。通过现场测试验证,该测量方法可有效地对板式无砟轨道层间结构位移进行精确测量,为我国高速铁路的运行安全监测提供技术支持。

Research on Measurement Method of Interlayer Structure Displacement in Ballastless Track Based on Machine Vision

In order to realize the all-weather monitoring of the slab-type ballastless interlayer separation, this paper proposes a non-contact inter-layer structure displacement measurement method based on machine vision. Compared with the traditional Canny edge detection algorithm, the Canny edge detection algorithm is better for its poor adaptability. Gaussian filter is replaced by adaptive median filter. A local threshold segmentation method based on block iteration is proposed to select the threshold value adaptively and detect the superiority of the improved algorithm by Gaussian fitting method. The optimum measurement parameters were selected by contrast analysis of the size of the mark, the intensity of the light source, the shooting distance and the shooting angle. The test results show that the maximum measurement error is 2.6% in the state of the marked point motion, which satisfies the measurement requirements of the track structure. Through field test and verification, the measurement method can effectively measure the structural displacement of the slab ballastless track and provide certain technical support for the operational safety monitoring of high-speed railways in China.