CN102628814A - Automatic detection method of steel rail light band abnormity based on digital image processing - Google Patents
Automatic detection method of steel rail light band abnormity based on digital image processing Download PDFInfo
- Publication number
- CN102628814A CN102628814A CN2012100470328A CN201210047032A CN102628814A CN 102628814 A CN102628814 A CN 102628814A CN 2012100470328 A CN2012100470328 A CN 2012100470328A CN 201210047032 A CN201210047032 A CN 201210047032A CN 102628814 A CN102628814 A CN 102628814A
- Authority
- CN
- China
- Prior art keywords
- rail
- image
- light belt
- edge
- straight line
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 26
- 238000012545 processing Methods 0.000 title claims abstract description 14
- 229910000831 Steel Inorganic materials 0.000 title claims description 3
- 239000010959 steel Substances 0.000 title claims description 3
- 239000000284 extract Substances 0.000 claims abstract description 17
- 238000000034 method Methods 0.000 claims abstract description 14
- 230000002159 abnormal effect Effects 0.000 claims abstract description 13
- 230000005856 abnormality Effects 0.000 claims abstract description 8
- 238000004458 analytical method Methods 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 12
- 238000005259 measurement Methods 0.000 claims description 6
- 230000007704 transition Effects 0.000 claims 2
- 238000001914 filtration Methods 0.000 claims 1
- 230000003287 optical effect Effects 0.000 claims 1
- 238000012360 testing method Methods 0.000 claims 1
- 238000003708 edge detection Methods 0.000 abstract description 4
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000003709 image segmentation Methods 0.000 abstract 1
- 238000003909 pattern recognition Methods 0.000 abstract 1
- 238000006243 chemical reaction Methods 0.000 description 7
- 230000009466 transformation Effects 0.000 description 7
- 230000007423 decrease Effects 0.000 description 4
- 238000012937 correction Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 229910052704 radon Inorganic materials 0.000 description 1
- SYUHGPGVQRZVTB-UHFFFAOYSA-N radon atom Chemical compound [Rn] SYUHGPGVQRZVTB-UHFFFAOYSA-N 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
本发明公开了一种基于数字图像处理的钢轨光带异常自动检测方法，采用数字图像处理技术分析相机拍摄的轨道图片并检测其钢轨光带是否发生异常；利用图像增强、边缘检测、直线检测等方法初步定位图像中钢轨区域；然后通过模式识别、图像分割、阈值处理等方法提取出钢轨顶面光带区域；最后，对所提取的钢轨光带区域进行统计、分析并识别出其是否发生异常。本发明可以高效、自动、智能的检测钢轨平稳性，有效地降低人力投入、减少检测时间，并保障检测的准确率，使得列车在高速运行时的安全得到有效保障。
The invention discloses a digital image processing-based automatic detection method for the abnormality of the rail light belt, which uses digital image processing technology to analyze the track pictures taken by the camera and detects whether the rail light belt is abnormal; uses image enhancement, edge detection, line detection, etc. The method preliminarily locates the rail area in the image; then extracts the light band area on the top surface of the rail through pattern recognition, image segmentation, threshold processing and other methods; finally, performs statistics and analysis on the extracted rail light band area and identifies whether it is abnormal . The invention can efficiently, automatically and intelligently detect the rail stability, effectively reduce manpower input and detection time, and ensure the accuracy of detection, so that the safety of the train at high speed is effectively guaranteed.
Description
技术领域 technical field
本发明涉及轨道平稳性能测量，尤其是高速铁路钢轨平稳性监测的采集钢轨图像并检测其中钢轨光带是否发生异常的方法。The invention relates to track stability performance measurement, in particular to a method for collecting rail images and detecting whether the rail light band is abnormal in the rail stability monitoring of high-speed railways.
背景技术 Background technique
随着高速铁路的发展，列车运行速度越来越快，对轨道平稳性的要求不断提高。然而，受到铁路修建、地理环境以及列车运行等因素的影响，轨道的平稳性难免会出现问题。由于轨道的平稳性会直接影响列车的运行安全，因此，对于轨道平稳性的检测，事关广大人民的生命财产安全。With the development of high-speed railways, trains run faster and faster, and the requirements for track stability continue to increase. However, affected by factors such as railway construction, geographical environment, and train operation, it is inevitable that there will be problems with the stability of the track. Since the stability of the track will directly affect the safety of the train, the detection of the track stability is related to the safety of people's lives and properties.
一般对轨道平稳性的检测是通过列车运行过程中留在钢轨上的光带进行判断的。钢轨光带是指当列车车轮在钢轨面滚动、滑动时，车轮轮缘与钢轨间相互作用，在钢轨上留下的亮痕。传统的钢轨光带检测是通过人工观察的方式进行判断的。由于高速列车在白天一直处于运行状态，在夜间停运，因此只能通过在夜间无车通行的情况下，使用探测灯对钢轨人工地逐一检测。全国高速铁路轨道在2012年将达到1.3万公里，要在短时间内通过人工检测的方法来判断钢轨光带是否发生异常极为困难。同时，由于人在夜间处于疲劳状态，加上光线等原因，容易对轨道平稳性的检测产生漏检和误检。这使得列车在高速运行时的安全难以得到有效保障。一种高效、自动、智能的钢轨平稳性检测方法成为必要。Generally, the detection of track stability is judged by the light band left on the rail during the train operation. The rail light strip refers to the bright marks left on the rail when the train wheel rolls and slides on the rail surface, the interaction between the wheel rim and the rail. The traditional rail light belt detection is judged by manual observation. Since high-speed trains are always in operation during the day and out of service at night, the rails can only be manually detected one by one by using detection lights when there is no traffic at night. The national high-speed railway track will reach 13,000 kilometers in 2012, and it is extremely difficult to judge whether there is an abnormality in the rail light belt through manual inspection in a short period of time. At the same time, because people are in a state of fatigue at night, coupled with light and other reasons, it is easy to miss and falsely detect the detection of track stability. This makes the safety of the train difficult to be effectively guaranteed when running at high speed. An efficient, automatic and intelligent rail stability detection method becomes necessary.
在人工检测的过程中，通常是观察钢轨顶面光带是否有明显的波纹、光带的宽度是否增大或减小来判断轨道是否平稳。在对钢轨顶面光带特征进行充分考虑，结合计算机视觉、数字图像处理技术，可以有效地降低人力投入、减少检测时间，并保障检测的准确率。In the process of manual detection, it is usually observed whether the light band on the top surface of the rail has obvious ripples and whether the width of the light band increases or decreases to judge whether the track is stable. Fully considering the characteristics of the light band on the top surface of the rail, combined with computer vision and digital image processing technology, can effectively reduce manpower input, reduce detection time, and ensure the accuracy of detection.
发明内容 Contents of the invention
鉴于现有技术的以上不足，本发明旨在提供一种利用计算机视觉、数字图像处理技术分析相机拍摄的轨道图片，并检测其钢轨光带是否发生异常的方法，使之克服现有技术的以上不足，高效、自动、智能地完成检测任务。In view of the above deficiencies in the prior art, the present invention aims to provide a method for utilizing computer vision and digital image processing technology to analyze the track pictures taken by the camera, and to detect whether the rail light belt is abnormal, so as to overcome the above deficiencies in the prior art. Insufficient, efficiently, automatically and intelligently complete the detection task.
本发明的目的通过如下手段来实现。The object of the present invention is achieved by the following means.
1)一种基于数字图像处理的钢轨光带异常自动检测方法，采用数字图像处理技术分析相机拍摄的轨道图片并检测其钢轨光带是否发生异常，其处理包含如下步骤：通过对大量正常钢轨表面区域进行实际测量，用刻度尺测量光带区域的宽度和两侧非光带区域的宽度，根据统计的大量数据得到光带区域的宽度阈值(宽度阈值为光带区域与非光带区域的比值)，然后根据方差计算每段钢轨区域光带边缘的波动值，得到光带区域的波纹阈值；1) An automatic detection method for rail light belt anomalies based on digital image processing. Digital image processing technology is used to analyze the track pictures taken by the camera and detect whether the rail light belt is abnormal. The processing includes the following steps: through a large number of normal rail surface The area is actually measured, and the width of the light band area and the width of the non-light band area on both sides are measured with a scale, and the width threshold of the light band area is obtained according to a large amount of statistical data (the width threshold is the ratio of the light band area to the non-light band area ), then calculate the fluctuation value of the edge of the light band in each section of rail area according to the variance, and obtain the ripple threshold in the light band area;
2)读取拍摄的钢轨图片，使用边缘检测的方法，提取出钢轨图像中的边缘纹理；2) Read the rail picture taken, and use the method of edge detection to extract the edge texture in the rail image;
3)对步骤2得到的边缘图像进行滤波，降低图像中的噪声干扰；3) Filter the edge image obtained in step 2 to reduce noise interference in the image;
4)若已知钢轨在图像当中的大致方向，则通过旋转使得钢轨在图像中接近垂直，直接转步骤6，否则，使用直线检测的方法找到步骤3得到的边缘纹理图像中最长的直线，用其方向作为图像中钢轨的方向；4) If the general direction of the rail in the image is known, rotate to make the rail close to vertical in the image, and go directly to step 6, otherwise, use the straight line detection method to find the longest straight line in the edge texture image obtained in step 3, Use its orientation as the orientation of the rail in the image;
5)若步骤4得到的钢轨方向接近垂直状态，则直接转步骤6，否则，对步骤3得到的边缘纹理图像进行旋转，使得钢轨方向接近垂直；5) If the rail direction obtained in step 4 is close to the vertical state, then directly go to step 6, otherwise, the edge texture image obtained in step 3 is rotated so that the rail direction is close to vertical;
6)从左到右提取边缘纹理图像中接近垂直方向上的直线，合并在垂直方向上过于接近的平行线，然后通过直线的长度以及其在原图对应区域的纹理特征、颜色特征，判断该直线是否为钢轨顶面两侧的边缘线，若是，则保存直线参数，最终，获得钢轨顶面两侧的边缘线。6) Extract straight lines close to the vertical direction in the edge texture image from left to right, merge the parallel lines that are too close in the vertical direction, and then judge the straight line by the length of the straight line and its texture features and color features in the corresponding area of the original image Whether it is the edge lines on both sides of the top surface of the rail, if so, save the straight line parameters, and finally, obtain the edge lines on both sides of the top surface of the rail.
7)提取边缘纹理图像中接近水平方向的直线，选取其中两条最长且非近邻的直线，作为图像中垂直于轨道的参考线；7) Extract straight lines close to the horizontal direction in the edge texture image, and select two longest and non-adjacent straight lines as reference lines perpendicular to the track in the image;
8)将提取出的钢轨顶面两侧边缘直线和接近水平方向的直线两两相交，得到四个交点；8) intersect the straight lines on both sides of the extracted rail top surface and the straight lines close to the horizontal direction to obtain four intersection points;
9)将步骤8所得四个交点对应的四边形映射为新的图片中的一个矩形，该矩形的四个顶点分别与步骤8所得四个交点一一对应；9) map the quadrilateral corresponding to the four intersections obtained in step 8 into a rectangle in the new picture, and the four vertices of the rectangle correspond to the four intersections obtained in step 8 respectively;
10)利用步骤8所得四个交点与步骤9矩形四顶点的对应关系，计算将四边形转换成矩形的转换矩阵；10) utilize step 8 gained four intersection points and the corresponding relation of step 9 rectangle four vertices, calculate quadrilateral conversion into the conversion matrix of rectangle;
11)利用步骤10所得转换矩阵将钢轨图像映射至步骤9所得新图，新图矩形两条垂直方向边之间区域为钢轨表面区域；11) Utilize the transformation matrix obtained in step 10 to map the rail image to the new figure obtained in step 9, and the area between the two vertical direction sides of the new figure rectangle is the surface area of the steel rail;
12)利用钢轨光带及非光带区域的颜色分布特征，结合阈值处理得到钢轨区域的光带区域和非光带区域，提取钢轨表面光带区域与非光带区域交界的边缘线，利用预先统计的波纹阈值分析光带是否存在波纹，如果存在波纹，则转步骤14；12) Using the color distribution characteristics of the rail light band and non-light band area, combined with threshold processing to obtain the light band area and non-light band area of the rail area, extract the edge line at the junction of the rail surface light band area and the non-light band area, and use the pre-statistic Whether there is ripple in the ripple threshold analysis light band, if there is ripple, then turn to step 14;
13)计算光带区域在钢轨表面所占宽度比例，利用预先统计的宽度阈值分析光带区域是否过宽或过窄，如果光带区域宽度正常，则提示正常，如果光带区域宽度异常，则转步骤14，转步骤2；13) Calculate the proportion of the width of the light band area on the rail surface, and use the pre-statistical width threshold to analyze whether the light band area is too wide or too narrow. If the width of the light band area is normal, it will prompt normal. If the width of the light band area is abnormal, then Go to step 14, go to step 2;
14)记录轨道表面异常，并记录异常点的物理位置信息，发出异常提示信号。14) Record the abnormality of the track surface, record the physical position information of the abnormal point, and send out an abnormality prompt signal.
采用本发明的方法，利用图像增强、边缘检测、直线检测等方法初步定位图像中钢轨区域；然后通过模式识别、图像分割、阈值处理等方法提取出钢轨顶面光带区域；最后，对所提取的钢轨光带区域进行统计、分析并识别出其是否发生异常。可以高效、自动、智能的检测钢轨平稳性，有效地降低人力投入、减少检测时间，并保障检测的准确率，使得列车在高速运行时的安全得到有效保障。Adopt the method of the present invention, utilize methods such as image enhancement, edge detection, straight line detection to initially locate the rail region in the image; Statistics, analysis and identification of abnormalities in the light strip area of the rail. It can efficiently, automatically and intelligently detect the rail stability, effectively reduce manpower input, reduce detection time, and ensure the accuracy of detection, so that the safety of the train at high speed is effectively guaranteed.
附图说明如下：The accompanying drawings are as follows:
图1是在钢轨检测中所需检测到的四种类型的光带。图中a表示钢轨顶面光带正常，b表示钢轨顶面光带减小，c表示钢轨顶面光带增大，d表示钢轨顶面光带有波纹。产生b的原因是列车在高速运行至“高包”处时，车体产生向上的加速度，车轮悬浮减载，钢轨上动态垂直力减小，车轮与钢轨顶面的接触面积减少，导致车轮在钢轨顶面上留下的光带减小。产生c的原因是列车高速运行至“低洼”处，钢轨上动态垂直力增加，车轮与钢轨顶面接触面积增大，导致车轮在钢轨顶面上留下的光带增大。产生d的原因是轨道发生松动或者轨道不平顺性，使得列车在运行过程中左右摇晃，导致车轮在钢轨顶面上留下的光带有波纹。Figure 1 shows the four types of light bands that need to be detected in rail detection. In the figure, a indicates that the light band on the top surface of the rail is normal, b indicates that the light band on the top surface of the rail decreases, c indicates that the light band on the top surface of the rail increases, and d indicates that the light band on the top surface of the rail has ripples. The reason for b is that when the train runs at high speed to the "high bag", the car body produces an upward acceleration, the wheel is suspended to reduce the load, the dynamic vertical force on the rail decreases, and the contact area between the wheel and the top surface of the rail decreases, resulting in a The light band left on the top surface of the rail is reduced. The reason for c is that the train runs to the "low-lying" place at high speed, the dynamic vertical force on the rail increases, and the contact area between the wheel and the top surface of the rail increases, resulting in an increase in the light band left by the wheel on the top surface of the rail. The cause of d is that the track is loose or the track is not smooth, which makes the train shake from side to side during operation, resulting in ripples in the light belt left by the wheels on the top surface of the rail.
图2是本发明设计的相机在轨道边沿轨道方向倾斜拍摄轨道图片的原理图。Fig. 2 is the schematic diagram of the camera designed in the present invention obliquely shooting track pictures along the track direction.
图3是本发明设计的相机在垂直于轨道正上方拍摄轨道图片的原理图。Fig. 3 is the principle diagram that the camera designed in the present invention shoots the track picture vertically above the track.
图4是本发明设计的在钢轨在图片中有一定倾斜和形变时，计算用于校正钢轨图片的转换矩阵的算法流程图。Fig. 4 is the algorithm flow chart for calculating the transformation matrix used to correct the rail picture when the rail has a certain inclination and deformation in the picture designed by the present invention.
图5是本发明利用原图和转换矩阵校正钢轨图片，然后提取并识别钢轨中光带区域是否发生异常的算法流程图。Fig. 5 is an algorithm flow chart of the present invention for correcting the rail picture by using the original image and the transformation matrix, and then extracting and identifying whether the light zone in the rail is abnormal.
具体实施方式 Detailed ways
下面结合附图和具体的实施方式对本发明作进一步的详细说明。The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.
本方法的具体步骤如下：The concrete steps of this method are as follows:
第一步，钢轨顶面光带宽度波动范围的测量。The first step is to measure the fluctuation range of the light band width on the top surface of the rail.
1)通过对大量正常钢轨表面区域进行实际测量，用刻度尺测量光带区域的宽度和两侧非光带区域的宽度，根据统计的大量数据得到光带区域的宽度阈值，然后根据方差计算每段钢轨区域中光带边缘的波动值，得到光带区域的波纹阈值；1) Through the actual measurement of a large number of normal rail surface areas, the width of the light band area and the width of the non-light band area on both sides are measured with a scale, and the width threshold of the light band area is obtained according to a large amount of statistical data, and then calculated according to the variance. The fluctuation value of the edge of the light band in the section rail area is obtained to obtain the ripple threshold of the light band area;
第二步，采集钢轨图像，提取钢轨顶面两侧边缘线和垂直于钢轨的参考，计算转换矩阵。The second step is to collect the rail image, extract the edge lines on both sides of the top surface of the rail and the reference perpendicular to the rail, and calculate the transformation matrix.
2)钢轨图像的采集建议，如图2和图3所示，图2是在钢轨两侧沿钢轨方向拍摄钢轨的图像，图3是在垂直方向上拍摄钢轨图像，如能将相机架设在轨检车上拍摄效果更佳，对采集的钢轨图像可以使用Canny、Sobel、Prewitt、Robert、小波、曲波、轮廓波等多种边缘提取算法，得到钢轨图像的边缘图像；2) Suggestions for the collection of rail images, as shown in Figure 2 and Figure 3, Figure 2 is the image of the rail taken along the direction of the rail on both sides of the rail, Figure 3 is the image of the rail taken in the vertical direction, if the camera can be set up on the rail The shooting effect on the inspection car is better, and various edge extraction algorithms such as Canny, Sobel, Prewitt, Robert, wavelet, curvelet, and contour wave can be used for the collected rail images to obtain the edge image of the rail image;
3)对步骤2得到的钢轨边缘图像，需要去除边缘图像中噪声干扰，可将图像进行平滑处理、连通性处理等，排除噪声点使得在边缘提取中距离较近但不连续的直线相互连接；3) For the rail edge image obtained in step 2, the noise interference in the edge image needs to be removed, and the image can be smoothed, connected, etc., and the noise points are excluded so that the closer but discontinuous straight lines are connected to each other in the edge extraction;
4)若已知钢轨在图像当中的大致方向，则通过旋转使得钢轨在图像中接近垂直，直接转步骤6，否则，可以使用Hough直线检测、Radon变换直线检测等，提取钢轨图像中较长的直线；4) If the general direction of the rail in the image is known, rotate the rail so that it is close to vertical in the image, and go directly to step 6, otherwise, use Hough line detection, Radon transformation line detection, etc. to extract the longer line in the rail image straight line;
5)对步骤4选取最长的直线即为钢轨方向，若钢轨方向接近垂直状态，则直接转步骤6，否则，将步骤4得到的钢轨图像进行旋转，使得钢轨方向接近垂直；5) Selecting the longest straight line in step 4 is the rail direction, if the rail direction is close to the vertical state, then directly go to step 6, otherwise, the rail image obtained in step 4 is rotated so that the rail direction is close to vertical;
6)从左到右提取边缘纹理图像中接近垂直方向上的直线，合并在垂直方向上过于接近的平行线，然后通过直线的长度以及其在原图对应区域的纹理特征、颜色特征，判断该直线是否为钢轨表面两侧的边缘线，若是，则保存直线参数，最终，获得钢轨表面两侧的边缘线；6) Extract straight lines close to the vertical direction in the edge texture image from left to right, merge the parallel lines that are too close in the vertical direction, and then judge the straight line by the length of the straight line and its texture features and color features in the corresponding area of the original image Whether it is the edge lines on both sides of the rail surface, if so, save the straight line parameters, and finally obtain the edge lines on both sides of the rail surface;
7)提取边缘纹理图像中接近水平方向的直线，选取其中两条最长且非近邻的直线，作为图像中垂直于轨道的参考线；7) Extract straight lines close to the horizontal direction in the edge texture image, and select two longest and non-adjacent straight lines as reference lines perpendicular to the track in the image;
8)计算钢轨两侧的直线和水平直线的交点，得到需要转换的四边形的顶点坐标，令其，左上点标为(x1，y1)，左下点坐标为(x2，y2)，右下点坐标为(x3，y3)，右上点坐标为(x4，y4)；8) Calculate the intersection point of the straight line on both sides of the rail and the horizontal straight line, and obtain the vertex coordinates of the quadrilateral that needs to be converted, so that the upper left point is marked as (x1, y1), the lower left point coordinates are (x2, y2), and the lower right point coordinates It is (x3, y3), and the coordinates of the upper right point are (x4, y4);
9)根据步骤8得到的四顶点构建新的矩形，令其左上角坐标为(x2，y1)，左下角坐标为(x2，y2)，右下角坐标为(x3，y2)，右上角坐标为(x3，y1)；9) Construct a new rectangle based on the four vertices obtained in step 8, let the coordinates of the upper left corner be (x2, y1), the coordinates of the lower left corner be (x2, y2), the coordinates of the lower right corner be (x3, y2), and the coordinates of the upper right corner be (x3, y1);
10)使用透视变换计算步骤8所得四边形转换为步骤9所得新矩形的转换矩阵；10) use perspective transformation calculation step 8 gained quadrilateral to be converted into the conversion matrix of step 9 gained new rectangle;
第三步，对钢轨图像进行校正，提取校正后的钢轨区域，然后提取钢轨光带区域，判断钢轨光带是否发生异常。The third step is to correct the rail image, extract the corrected rail area, and then extract the rail light band area to determine whether the rail light band is abnormal.
11)利用步骤10所得转换矩阵计算钢轨矫正后的钢轨图像映射至步骤9所得新图，新图矩形两条垂直方向线段之间区域为钢轨表面区域；11) Utilize the conversion matrix obtained in step 10 to calculate the rail image after the rail correction is mapped to the new figure obtained in step 9, and the area between the two vertical line segments of the new figure rectangle is the surface area of the rail;
12)利用钢轨光带及非光带区域的颜色分布特征，结合阈值处理得到钢轨区域的光带区域和非光带区域，提取钢轨表面光带区域与非光带区域交界的边缘线，利用步骤1预先统计的波纹阈值分析光带是否存在波纹，如果存在波纹，则转步骤14；12) Utilize the color distribution characteristics of the rail light band and the non-light band area, combine the threshold value processing to obtain the light band area and the non-light band area of the rail area, extract the edge line at the junction of the rail surface light band area and the non-light band area, and use step 1 The pre-statistical ripple threshold analyzes whether there are ripples in the light band, and if there are ripples, go to step 14;
13)计算光带区域在钢轨表面所占宽度比例，利用预先统计的宽度阈值分析光带区域是否过宽或过窄，如果光带区域宽度正常，则提示正常，如果光带区域宽度异常，则转步骤14，转步骤2；13) Calculate the proportion of the width of the light band area on the rail surface, and use the pre-statistical width threshold to analyze whether the light band area is too wide or too narrow. If the width of the light band area is normal, it will prompt normal. If the width of the light band area is abnormal, then Go to step 14, go to step 2;
14)记录轨道表面异常，并记录异常点的物理位置信息，发出异常提示信息。14) Record the track surface abnormality, and record the physical location information of the abnormal point, and issue an abnormality prompt message.
实施例Example
以下是本发明的实例步骤说明：The following is an example step description of the present invention:
1)通过人工方式对大量正常钢轨表面区域进行实际测量，用刻度尺测量光带区域的宽度和两侧非光带区域的宽度，根据统计的大量数据得到光带区域的宽度阈值(根据实际测量，取得宽度阈值为0.5到0.6之间)，然后根据方差计算每段钢轨区域中光带边缘的波动值，得到光带区域的波纹阈值(根据实际测量，取得波纹阈值小于0.01)；1) The actual measurement of a large number of normal rail surface areas is carried out manually, and the width of the light band area and the width of the non-light band areas on both sides are measured with a scale, and the width threshold of the light band area is obtained according to a large amount of statistical data (according to actual measurement , obtain the width threshold value between 0.5 and 0.6), then calculate the fluctuation value of the light band edge in each section of rail area according to the variance, and obtain the ripple threshold value of the light band area (according to actual measurement, the ripple threshold value is less than 0.01);
2)钢轨图像的采集如图2所示，使用Canny边缘检测得到钢轨图像的边缘图像；2) The acquisition of the rail image is shown in Figure 2, using Canny edge detection to obtain the edge image of the rail image;
3)对步骤2得到的钢轨边缘图像，使用Smooth平滑处理(即对图像进行核大小为3*3的高斯卷积)排除噪声点使得在边缘提取中距离较近但不连续的直线相互连接；3) For the rail edge image obtained in step 2, use Smooth smoothing (i.e., carry out Gaussian convolution with a kernel size of 3*3 on the image) to eliminate noise points so that the closer but discontinuous straight lines are connected to each other in the edge extraction;
4)若已知钢轨在图像当中的大致方向，则通过旋转使得钢轨在图像中接近垂直，直接转步骤6，否则，可以使用Hough直线检测提取钢轨图像中较长的直线；4) If the general direction of the rail in the image is known, then rotate to make the rail close to vertical in the image, and go to step 6 directly, otherwise, use Hough straight line detection to extract the longer straight line in the rail image;
5)对步骤4选取最长的直线即为钢轨方向，若钢轨方向接近垂直状态，则直接转步骤6，否则，将步骤4得到的钢轨图像进行旋转，使得钢轨方向接近垂直；5) Selecting the longest straight line in step 4 is the rail direction, if the rail direction is close to the vertical state, then directly go to step 6, otherwise, the rail image obtained in step 4 is rotated so that the rail direction is close to vertical;
6)从左到右提取边缘纹理图像中接近垂直方向上的直线，合并在垂直方向上过于接近的平行线，然后通过直线的长度以及其在原图对应区域的颜色特征，判断该直线是否为钢轨表面两侧的边缘线，若是，则保存直线参数，最终，获得钢轨表面两侧的边缘线；6) Extract the straight lines close to the vertical direction in the edge texture image from left to right, merge the parallel lines that are too close in the vertical direction, and then judge whether the straight line is a rail by the length of the straight line and its color feature in the corresponding area of the original image The edge lines on both sides of the surface, if so, save the straight line parameters, and finally, obtain the edge lines on both sides of the rail surface;
7)提取边缘纹理图像中接近水平方向的直线，选取其中两条最长且非近邻的直线，作为图像中垂直于轨道的参考线；7) Extract straight lines close to the horizontal direction in the edge texture image, and select two longest and non-adjacent straight lines as reference lines perpendicular to the track in the image;
8)计算钢轨两侧的直线和水平直线的交点，得到需要转换的四边形的顶点坐标，令其，左上点标为(x1，y1)，左下点坐标为(x2，y2)，右下点坐标为(x3，y3)，右上点坐标为(x4，y4)；8) Calculate the intersection point of the straight line on both sides of the rail and the horizontal straight line, and obtain the vertex coordinates of the quadrilateral that needs to be converted, so that the upper left point is marked as (x1, y1), the lower left point coordinates are (x2, y2), and the lower right point coordinates It is (x3, y3), and the coordinates of the upper right point are (x4, y4);
9)根据步骤8得到的四顶点构建新的矩形，令其左上角坐标为(x2，y1)，左下角坐标为(x2，y2)，右下角坐标为(x3，y2)，右上角坐标为(x3，y1)；9) Construct a new rectangle based on the four vertices obtained in step 8, let the coordinates of the upper left corner be (x2, y1), the coordinates of the lower left corner be (x2, y2), the coordinates of the lower right corner be (x3, y2), and the coordinates of the upper right corner be (x3, y1);
10)使用透视变换计算步骤8所得四边形转换为步骤9所得新矩形的转换矩阵；10) use perspective transformation calculation step 8 gained quadrilateral to be converted into the conversion matrix of step 9 gained new rectangle;
11)利用步骤10所得转换矩阵计算钢轨矫正后的钢轨图像映射至步骤9所得新图，新图矩形两条垂直方向线段之间区域为钢轨表面区域；11) Utilize the conversion matrix obtained in step 10 to calculate the rail image after the rail correction is mapped to the new figure obtained in step 9, and the area between the two vertical line segments of the new figure rectangle is the surface area of the rail;
12)利用步骤10所得转换矩阵计算钢轨矫正后的钢轨图像映射至步骤9所得新图，新图矩形两条垂直方向线段之间区域为钢轨表面区域；12) Utilize the conversion matrix obtained in step 10 to calculate the rail image after rail correction to map to the new figure obtained in step 9, and the area between the two vertical line segments of the new figure rectangle is the surface area of the rail;
13)计算光带区域在钢轨表面所占宽度比例，利用预先统计的宽度阈值分析光带区域是否过宽或过窄，如果光带区域宽度正常，则提示正常，如果光带区域宽度异常，则转步骤14，转步骤2；13) Calculate the proportion of the width of the light band area on the rail surface, and use the pre-statistical width threshold to analyze whether the light band area is too wide or too narrow. If the width of the light band area is normal, it will prompt normal. If the width of the light band area is abnormal, then Go to step 14, go to step 2;
记录轨道表面异常，并记录异常点的物理位置信息，发出异常提示信息。Record track surface anomalies, record physical location information of anomalies, and issue anomalies prompt information.
Claims (1)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201210047032 CN102628814B (en) | 2012-02-28 | 2012-02-28 | Automatic detection method of steel rail light band abnormity based on digital image processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201210047032 CN102628814B (en) | 2012-02-28 | 2012-02-28 | Automatic detection method of steel rail light band abnormity based on digital image processing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102628814A true CN102628814A (en) | 2012-08-08 |
| CN102628814B CN102628814B (en) | 2013-12-18 |
Family
ID=46587113
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201210047032 Expired - Fee Related CN102628814B (en) | 2012-02-28 | 2012-02-28 | Automatic detection method of steel rail light band abnormity based on digital image processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102628814B (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104197842A (en) * | 2014-09-15 | 2014-12-10 | 常州宝仪机电设备有限公司 | Image type plate width measurement and control system |
| CN104713486A (en) * | 2015-03-20 | 2015-06-17 | 西安交通大学 | A device and method for on-line detection of the width of gasshang products |
| CN104792270A (en) * | 2015-04-16 | 2015-07-22 | 石家庄铁路职业技术学院 | Deep foundation pit detection method based on white-light digital image frequency domain analysis method |
| CN105466941A (en) * | 2015-11-27 | 2016-04-06 | 中国铁道科学研究院 | Steel rail detection method and equipment based on light band image of steel rail |
| CN105531581A (en) * | 2013-09-10 | 2016-04-27 | 蒂森克虏伯钢铁欧洲股份公司 | Method and device for testing an inspection system for detecting surface defects |
| CN106355185A (en) * | 2016-08-30 | 2017-01-25 | 兰州交通大学 | Method for rapidly extracting steel rail surface area under condition of vibration |
| CN106428558A (en) * | 2016-11-28 | 2017-02-22 | 北京交通大学 | Rail comprehensive inspection method based on air-rail double-purpose unmanned aerial vehicle |
| CN106494611A (en) * | 2016-11-28 | 2017-03-15 | 北京交通大学 | A kind of dual-purpose patrol unmanned machine of empty rail |
| CN107194921A (en) * | 2017-05-19 | 2017-09-22 | 华夏高铁技术有限公司 | The automatic identifying method and device of rail rail level light belt |
| CN107238605A (en) * | 2017-06-08 | 2017-10-10 | 广东嘉铭智能科技有限公司 | A kind of cylindric outer surface of workpiece defect inspection method and device |
| CN107264570A (en) * | 2017-07-25 | 2017-10-20 | 西南交通大学 | steel rail light band distribution detecting device and method |
| CN108510475A (en) * | 2018-03-09 | 2018-09-07 | 南京索聚医疗科技有限公司 | The measurement method and system of muscle tendon knot in a kind of muscle continuous ultrasound image |
| CN109117757A (en) * | 2018-07-27 | 2019-01-01 | 四川大学 | A kind of method of drag-line in extraction Aerial Images |
| CN111105409A (en) * | 2019-12-27 | 2020-05-05 | 中国铁道科学研究院集团有限公司 | Automatic detection method, equipment and system for steel rail light band based on image processing |
| CN113112474A (en) * | 2021-04-12 | 2021-07-13 | 东南大学 | Rail positioning method for turnout and non-turnout track images |
| CN114166855A (en) * | 2022-02-11 | 2022-03-11 | 南京邮电大学 | Real-time rail defect detection method |
| CN114384078A (en) * | 2021-12-31 | 2022-04-22 | 武汉大学 | Method and equipment for location detection of rail corrugation damage |
| CN116109638A (en) * | 2023-04-13 | 2023-05-12 | 中铁四局集团有限公司 | Method and system for detecting rail breakage |
| CN116630351A (en) * | 2023-05-26 | 2023-08-22 | 广州市交通设计研究院有限公司 | A method, device, terminal equipment and storage medium for generating a core histogram |
| CN116703847A (en) * | 2023-05-31 | 2023-09-05 | 西南交通大学 | A method, system, device, and storage medium for identifying light strips of a turnout rail |
| CN117253066A (en) * | 2023-11-20 | 2023-12-19 | 西南交通大学 | Rail surface state identification method, device, equipment and readable storage medium |
| CN118365594A (en) * | 2024-04-02 | 2024-07-19 | 青岛碳峰智能科技工程有限公司 | Door and window detection method and system based on image recognition |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5621645A (en) * | 1995-01-24 | 1997-04-15 | Minnesota Mining And Manufacturing Company | Automated lane definition for machine vision traffic detector |
| US5937079A (en) * | 1996-09-05 | 1999-08-10 | Daimler-Benz Ag | Method for stereo image object detection |
| CN1645037A (en) * | 2004-12-31 | 2005-07-27 | 天津大学 | Partitioned pointing method for three-dimensional scanning measurement system by light band method |
| CN101620732A (en) * | 2009-07-17 | 2010-01-06 | 南京航空航天大学 | Visual detection method of road driving line |
| US20100079590A1 (en) * | 2007-01-23 | 2010-04-01 | Valeo Schalter Und Sensoren Gmbh | Method and system for video-based road characterization, lane detection and departure prevention |
-
2012
- 2012-02-28 CN CN 201210047032 patent/CN102628814B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5621645A (en) * | 1995-01-24 | 1997-04-15 | Minnesota Mining And Manufacturing Company | Automated lane definition for machine vision traffic detector |
| US5937079A (en) * | 1996-09-05 | 1999-08-10 | Daimler-Benz Ag | Method for stereo image object detection |
| CN1645037A (en) * | 2004-12-31 | 2005-07-27 | 天津大学 | Partitioned pointing method for three-dimensional scanning measurement system by light band method |
| US20100079590A1 (en) * | 2007-01-23 | 2010-04-01 | Valeo Schalter Und Sensoren Gmbh | Method and system for video-based road characterization, lane detection and departure prevention |
| CN101620732A (en) * | 2009-07-17 | 2010-01-06 | 南京航空航天大学 | Visual detection method of road driving line |
Non-Patent Citations (1)
| Title |
|---|
| 靳文瑞等: "《近红外钢轨磨耗检测中的图像处理》", 《应用光学》 * |
Cited By (35)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105531581A (en) * | 2013-09-10 | 2016-04-27 | 蒂森克虏伯钢铁欧洲股份公司 | Method and device for testing an inspection system for detecting surface defects |
| CN105531581B (en) * | 2013-09-10 | 2019-07-12 | 蒂森克虏伯钢铁欧洲股份公司 | The method and apparatus tested to the inspection system of surface defect for identification |
| CN104197842A (en) * | 2014-09-15 | 2014-12-10 | 常州宝仪机电设备有限公司 | Image type plate width measurement and control system |
| CN104713486B (en) * | 2015-03-20 | 2017-08-15 | 西安交通大学 | Device and method for online detection of width of palm folding product |
| CN104713486A (en) * | 2015-03-20 | 2015-06-17 | 西安交通大学 | A device and method for on-line detection of the width of gasshang products |
| CN104792270A (en) * | 2015-04-16 | 2015-07-22 | 石家庄铁路职业技术学院 | Deep foundation pit detection method based on white-light digital image frequency domain analysis method |
| CN105466941A (en) * | 2015-11-27 | 2016-04-06 | 中国铁道科学研究院 | Steel rail detection method and equipment based on light band image of steel rail |
| CN106355185A (en) * | 2016-08-30 | 2017-01-25 | 兰州交通大学 | Method for rapidly extracting steel rail surface area under condition of vibration |
| CN106355185B (en) * | 2016-08-30 | 2019-07-16 | 兰州交通大学 | A method for rapid extraction of rail surface area under vibration conditions |
| CN106494611A (en) * | 2016-11-28 | 2017-03-15 | 北京交通大学 | A kind of dual-purpose patrol unmanned machine of empty rail |
| CN106428558A (en) * | 2016-11-28 | 2017-02-22 | 北京交通大学 | Rail comprehensive inspection method based on air-rail double-purpose unmanned aerial vehicle |
| CN106428558B (en) * | 2016-11-28 | 2018-06-01 | 北京交通大学 | A kind of track synthesis method for inspecting based on the dual-purpose unmanned plane of sky-rail |
| CN107194921B (en) * | 2017-05-19 | 2020-04-07 | 华夏高铁技术有限公司 | Automatic identification method and device for rail surface light band of steel rail |
| CN107194921A (en) * | 2017-05-19 | 2017-09-22 | 华夏高铁技术有限公司 | The automatic identifying method and device of rail rail level light belt |
| CN107238605A (en) * | 2017-06-08 | 2017-10-10 | 广东嘉铭智能科技有限公司 | A kind of cylindric outer surface of workpiece defect inspection method and device |
| CN107238605B (en) * | 2017-06-08 | 2021-07-02 | 广东嘉铭智能科技有限公司 | A method and device for detecting defects on the outer surface of a cylindrical workpiece |
| CN107264570B (en) * | 2017-07-25 | 2019-07-05 | 西南交通大学 | Rail light strip distribution detection device and method |
| CN107264570A (en) * | 2017-07-25 | 2017-10-20 | 西南交通大学 | steel rail light band distribution detecting device and method |
| CN108510475A (en) * | 2018-03-09 | 2018-09-07 | 南京索聚医疗科技有限公司 | The measurement method and system of muscle tendon knot in a kind of muscle continuous ultrasound image |
| CN109117757A (en) * | 2018-07-27 | 2019-01-01 | 四川大学 | A kind of method of drag-line in extraction Aerial Images |
| CN109117757B (en) * | 2018-07-27 | 2022-02-22 | 四川大学 | A method for extracting cables in aerial images |
| CN111105409A (en) * | 2019-12-27 | 2020-05-05 | 中国铁道科学研究院集团有限公司 | Automatic detection method, equipment and system for steel rail light band based on image processing |
| CN111105409B (en) * | 2019-12-27 | 2023-08-08 | 中国铁道科学研究院集团有限公司 | Method, device and system for automatic detection of rail light strips based on image processing |
| CN113112474A (en) * | 2021-04-12 | 2021-07-13 | 东南大学 | Rail positioning method for turnout and non-turnout track images |
| CN114384078B (en) * | 2021-12-31 | 2023-11-24 | 武汉大学 | Method and equipment for detecting wave abrasion damage of steel rail in positioning mode |
| CN114384078A (en) * | 2021-12-31 | 2022-04-22 | 武汉大学 | Method and equipment for location detection of rail corrugation damage |
| CN114166855A (en) * | 2022-02-11 | 2022-03-11 | 南京邮电大学 | Real-time rail defect detection method |
| CN116109638A (en) * | 2023-04-13 | 2023-05-12 | 中铁四局集团有限公司 | Method and system for detecting rail breakage |
| CN116109638B (en) * | 2023-04-13 | 2023-07-04 | 中铁四局集团有限公司 | Rail break detection method and system |
| CN116630351A (en) * | 2023-05-26 | 2023-08-22 | 广州市交通设计研究院有限公司 | A method, device, terminal equipment and storage medium for generating a core histogram |
| CN116703847A (en) * | 2023-05-31 | 2023-09-05 | 西南交通大学 | A method, system, device, and storage medium for identifying light strips of a turnout rail |
| CN117253066A (en) * | 2023-11-20 | 2023-12-19 | 西南交通大学 | Rail surface state identification method, device, equipment and readable storage medium |
| CN117253066B (en) * | 2023-11-20 | 2024-02-27 | 西南交通大学 | Methods, devices, equipment and readable storage media for identifying rail surface conditions |
| CN118365594A (en) * | 2024-04-02 | 2024-07-19 | 青岛碳峰智能科技工程有限公司 | Door and window detection method and system based on image recognition |
| CN118365594B (en) * | 2024-04-02 | 2025-01-21 | 青岛碳峰智能科技工程有限公司 | A door and window detection method and system based on image recognition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102628814B (en) | 2013-12-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102628814B (en) | Automatic detection method of steel rail light band abnormity based on digital image processing | |
| CN104112370B (en) | Parking lot based on monitoring image intelligent car position recognition methods and system | |
| CN103488976B (en) | Distance measurement method based on stop mark real-time detection during intelligent driving | |
| CN107025432B (en) | A kind of efficient lane detection tracking and system | |
| CN103217111B (en) | A kind of non-contact contact line geometric parameter detection method | |
| CN107203990B (en) | A Label Damage Detection Method Based on Template Matching and Image Quality Assessment | |
| CN103955923B (en) | A kind of quickly pavement disease detection method based on image | |
| CN102788803B (en) | On-line high-speed detection system of train wheel set tread and wheel rim and detection method thereof | |
| CN107618533A (en) | A kind of machine vision detection device and method of the discrete defect of Rail Surface | |
| CN102999749B (en) | Based on the securing band violation event intelligent detecting method of Face datection | |
| CN103051872B (en) | Based on the belt deflection detection method of Edge extraction | |
| CN107588733A (en) | A kind of abrasion of pantograph pan on-line measuring device and method based on image | |
| CN105260713A (en) | Method and device for detecting lane line | |
| CN108281041A (en) | A kind of parking space's detection method blended based on ultrasonic wave and visual sensor | |
| CN103593649A (en) | Lane line detection method for lane departure early warning | |
| CN106052575A (en) | Pantograph carbon slide plate wear online detection method based on train high speed running | |
| CN103500322A (en) | Automatic lane line identification method based on low-altitude aerial images | |
| CN108830841A (en) | The calculation method of laser image characteristic quantity selection threshold value | |
| CN107180416A (en) | Train wheel tread image deformity correction method and system | |
| CN105718916A (en) | Lane line detection method based on Hough transform | |
| CN107727658A (en) | Pantograph crackle on-line measuring device and method based on image procossing | |
| CN102542799A (en) | Line acquisition video vehicle detector based on pavement marker and detecting method thereof | |
| CN108830822B (en) | Pantograph carbon skateboard abrasion identification method based on improved Canny operator | |
| CN110047070A (en) | A kind of identification method and system of track wear degree | |
| CN108399403A (en) | A kind of vehicle distance detecting method calculated based on car plate size |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20131218 Termination date: 20180228 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |