Steel rail abrasion measuring method and device

Through the checkerboard method and ICP algorithm, the problems of low efficiency and low accuracy in rail wear detection are solved, high-precision rail wear detection is realized, point cloud data registration process is simplified, and detection efficiency and accuracy are improved.

CN119090971AInactive Publication Date: 2024-12-06SHANGHAI INST OF TECH
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411051710.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-01
Publication Date
2024-12-06
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing rail wear detection methods are inefficient and have low accuracy, complex image processing, and difficult point cloud data registration, especially in dynamic environments or long-distance scanning, and the traditional equipment is complex to operate, requiring professional operation.

Method used

The checkerboard method is used to perform dual-target determination, the internal and external parameters of the camera are obtained, the laser line profile is obtained through light plane calibration and image processing, the point cloud data is registered using the ICP algorithm, and the rail wear value is calculated.

Benefits of technology

Simplify point cloud profile, improve matching accuracy, reduce alignment errors, realize high-precision rail wear detection, and improve detection efficiency and accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119090971A_ABST
    Figure CN119090971A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of track abrasion detection, in particular to a steel rail abrasion measuring method and device.The method comprises the steps that binocular calibration is conducted through a checkerboard method, and internal and external parameters of a left camera and a right camera are obtained; obtaining a light plane equation of the light plane in a left and right camera coordinate system through light plane calibration; image processing: controlling a left camera and a right camera through a computer to acquire a plurality of groups of pictures including laser line contours containing steel rail information, and performing distortion correction processing on the acquired original pictures; acquiring point cloud data and registering the point cloud data; and calculating the steel rail abrasion value. The method has the advantages that the point clouds of the left camera and the right camera are registered through the ICP algorithm, the point clouds can be accurately aligned, dislocation caused by different positions and angles of the cameras is eliminated, the fused point clouds contain more detail information, the integrity and accuracy of a three-dimensional model are improved, and the accuracy of the three-dimensional model is improved. And high-precision matching between the simplified point cloud contour and the standard contour is ensured.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of rail wear detection, in particular to a rail wear measurement method and device. Background Art

[0002] Existing rail wear detection methods mainly rely on manual inspection and traditional measuring equipment, which have problems such as low efficiency, low accuracy, and complex operation. However, laser scanning technology can quickly obtain three-dimensional data of the rail surface, which can theoretically greatly improve the efficiency and accuracy of detection. However, there are still some technical difficulties in image processing and data analysis in existing systems. First, the images obtained by laser scanning often contain noise and distortion, and complex image preprocessing steps are required to improve data quality. Secondly, the registration of point cloud data is a technical challenge, especially in dynamic environments or when scanning over long distances, the alignment and fusion of point cloud data requires precise algorithms and computing resources. Moreover, most existing systems use single-view shooting and recording, which has technical difficulties such as complex image processing and difficult registration of point cloud data. In addition, although traditional measuring equipment provides automation to a certain extent, their accuracy is usually limited by equipment performance and environmental conditions, and the operation process is complicated, requiring professionals to operate and analyze the data. The amount of point cloud data is huge, and efficient data compression and indexing technology is required to optimize the storage and retrieval process. Summary of the invention

[0003] In view of the above problems or problems existing in the prior art, the present invention is proposed.

[0004] Therefore, the object of the present invention is to provide a rail wear measurement method, which can simplify the point cloud profile, improve the matching accuracy, and reduce the alignment error.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a rail wear measurement method, which includes performing binocular positioning by a chessboard method to obtain internal and external parameters of left and right cameras;

[0006] The light plane equation of the light plane in the left and right camera coordinate systems is obtained through light plane calibration;

[0007] Image processing: the left and right cameras are controlled by a computer to collect multiple sets of images including the laser line profile containing rail information, and the original images are subjected to distortion correction processing;

[0008] Acquire point cloud data and register the point cloud data;

[0009] Calculate rail wear value.

[0010] As a preferred embodiment of the rail wear measurement method of the present invention, the checkerboard method is a checkerboard pattern composed of black and white squares, usually 9x6 or 12x9;

[0011] The internal and external parameters of the left and right cameras include the internal and external parameter matrices of the cameras on both sides. The expressions of the internal and external parameter matrices use the calibrateCamera function and stereoCalibrate function respectively. The expressions are as follows:

[0012]

[0013] Among them, f x 、f y is the focal length, c x 、c y is the principal point coordinate, R is the rotation matrix between cameras, T is the translation vector, and r is the distortion coefficient.

[0014] As a preferred solution of the rail wear measurement method of the present invention, wherein: the light plane calibration method includes calculating the light plane by the light plane equation, and calculating the light plane equation by point cloud fitting;

[0015] The light plane equation is as follows:

[0016] ax+by+cz+d=0

[0017] Among them, (a, b, c) is the normal vector, and d is the distance from the plane to the origin;

[0018] Point cloud fitting involves fitting points on the light plane using the least squares method and calculating the light plane equation, which is as follows:

[0019] ax+by+cz+d=0

[0020] Among them, (a, b, c) is the normal vector and d is a constant term.

[0021] As a preferred solution of the rail wear measurement method of the present invention, the image processing includes image preprocessing and extracting the center line of the laser line by Steger algorithm;

[0022] Image preprocessing methods include iterative thresholding and binarization, as well as ROI template matching;

[0023] The thresholding method and binarization include the initial selection of the threshold and iterative adjustment of the threshold until the image segmentation effect is satisfactory; the final threshold is applied to binarize the image. The calculation formula of the thresholding method and binarization is:

[0024]

[0025] Where T is the threshold, I filtered (x, y) is the image after Gaussian filtering, I binary (x, y) is a binary image;

[0026] The ROI template matching method includes further extracting the laser line area through template matching, removing redundant interference information, and performing template matching through sliding windows and correlation calculation:

[0027]

[0028] Where T is the template image, ω and h are the width and height of the template respectively;

[0029] The Steger algorithm calculates the centerline intensity of each contour point and selects the position with the highest intensity as the corresponding point of the centerline to form the centerline of the laser line. The calculation formula is as follows:

[0030]

[0031] Among them, C(x, y) is the centerline intensity at point (x, y), I(x+u, y+v) is the image pixel value, G(u, v) is the Gaussian function used to weight the surrounding pixels, w is the normalization factor, and k determines the size of the window.

[0032] As a preferred solution of the rail wear measurement method of the present invention, the point cloud data acquisition includes obtaining the three-dimensional coordinates (X, Y, Z) of the point cloud data according to the pixel points (u, v) and the depth information Z of the camera image coordinate system, and the calculation formula is as follows:

[0033]

[0034] Among them, (c x , c y ) is the optical center coordinate, f x , f y is the focal length;

[0035] Point cloud data registration includes registering and fusing the point cloud contours of the left and right cameras through the ICP algorithm. The calculation steps include solving the rigid body transformation R and t that minimizes the mean square error between the source point cloud and the target point cloud. The formula is:

[0036]

[0037] By taking the derivative of the above formula and setting it to zero, we can get the following system of equations:

[0038]

[0039] Among them, R is the rotation matrix, t is the translation vector, and are the centroids of P and Q respectively, and U and V come from the covariance matrix The singular value decomposition of H=U∑V T , take R = I (unit matrix), t = 0 (zero vector);

[0040] The calculation step also includes, for the fused point cloud, using the local average method to reduce noise and smooth the point cloud contour, including calculating the average value of all noise points in the selected field point, that is, for each point P in the point cloud i (x i ,y i , z i ), find all the neighborhood points P within its neighborhood radius j (x j ,y j , z j ) is a point that satisfies the following conditions:

[0041]

[0042] The average value is calculated as follows:

[0043]

[0044] Where N is the number of neighborhood points, is the smoothed coordinate value.

[0045] As a preferred solution of the rail wear measurement method of the present invention, the rail wear calculation method includes: performing polyline fitting according to the point cloud data to generate polyline functions f(x1) and f(x2) of the rail head, and obtaining the rail wear value by calculating the difference between the two functions at the vertical wear position and the side wear position, wherein the function form of the polyline fitting is:

[0046]

[0047] Among them, a i and b i are the slope and intercept of the fitted line segment, is the interval range of each line segment.

[0048] In order to solve the above technical problems, the present invention also provides the following technical solutions: A rail wear measuring device comprises a bearing unit, including a carrying component and a supporting component arranged at the lower side of the carrying component;

[0049] The measuring unit comprises a measuring component arranged on the upper side of the object-carrying component.

[0050] As a preferred solution of the rail wear measurement method of the present invention, the object-carrying assembly includes an object-carrying platform, and an installation groove is provided on the inner side of the object-carrying platform.

[0051] As a preferred solution of the rail wear measuring device of the present invention, the support assembly includes a fixed bracket arranged on the lower side of the loading platform, and a fixed foot arranged on the outer side of the fixed bracket.

[0052] As a preferred solution of the rail wear measuring device of the present invention, the measuring component includes a line laser arranged on the upper side of the loading platform, two industrial cameras arranged on both sides of the line laser, and a computer connected to the industrial cameras.

[0053] Beneficial effects of the present invention: The present invention registers the point clouds of the left and right cameras through the ICP algorithm, and can obtain a unified point cloud data set, which reflects the three-dimensional form of the same object observed from different perspectives. The fused points enable the point clouds captured by the left and right cameras to be accurately aligned, eliminating the misalignment caused by different camera positions and angles, and the fused point cloud contains more detailed information because it integrates data from multiple perspectives. The registered point cloud can cover more parts of the object and improve the integrity and accuracy of the three-dimensional model. ICP also reduces errors through the initial alignment of coarse registration, and then through the optimization steps of fine registration, ensures high-precision matching between the simplified point cloud contour and the standard contour. This precise alignment is of great significance for subsequent analysis and application. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0055] Figure 1 This is a system detection flow chart of the rail wear measurement method.

[0056] Figure 2 Schematic diagram of the image processing process of the rail wear measurement method.

[0057] Figure 3 The original point cloud data of the rail taken by the left camera and the right camera for the rail wear measurement method.

[0058] Figure 4 Schematic diagram of point cloud data fusion for rail wear measurement method.

[0059] Figure 5 This is a comparison diagram of the effect of point cloud rough alignment and noise reduction and smoothing processing of the rail wear measurement method and the ideal contour model of the standard rail.

[0060] Figure 6 Schematic diagram of precise registration of point cloud data and standard rail profile for rail wear measurement method.

[0061] Figure 7 Schematic diagram of point cloud data processing using multi-segment line fitting technology for rail wear measurement.

[0062] Figure 8 Rail wear calculation diagram for rail wear measurement method.

[0063] Fig. 9 This is a schematic diagram of the overall structure of the rail wear measuring device. DETAILED DESCRIPTION

[0064] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0065] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0066] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0067] Example 1

[0068] Reference Figures 1 to 8 , which is the first embodiment of the present invention, provides a rail wear measurement method, which can simplify the point cloud contour, improve matching accuracy, and reduce alignment errors.

[0069] Specifically, the checkerboard method is used to perform binocular positioning to obtain the internal and external parameters of the left and right cameras;

[0070] The light plane equation of the light plane in the left and right camera coordinate systems is obtained through light plane calibration;

[0071] Image processing: the left and right cameras are controlled by a computer to collect multiple sets of images including the laser line profile containing rail information, and the original images are subjected to distortion correction processing;

[0072] Acquire point cloud data and register the point cloud data;

[0073] Calculate rail wear value.

[0074] Further, the checkerboard method is a checkerboard pattern consisting of black and white squares, usually 9x6 or 12x9;

[0075] The internal and external parameters of the left and right cameras include the internal and external parameter matrices of the cameras on both sides. The internal and external parameter matrices use the calibrateCamera function and stereoCalibrate function respectively. The expressions are as follows:

[0076]

[0077] Among them, f x 、f y is the focal length, c x 、c y is the principal point coordinate, R is the rotation matrix between cameras, T is the translation vector, and r is the distortion coefficient.

[0078] It should be noted that the specific steps of performing dual-target positioning by the chessboard method also include:

[0079] Printing checkerboard pattern: The checkerboard pattern should have enough black and white squares, usually 9x6 or 12x9;

[0080] Fixed calibration board: Fix the checkerboard pattern on a flat surface;

[0081] Capture multiple images simultaneously: The left and right cameras simultaneously capture multiple checkerboard images at different angles to ensure coverage of different viewing angles and distances;

[0082] Extract corner points: The function used to detect checkerboard corner points in the image;

[0083] Camera intrinsic parameter matrix calculation: Through the above function, the intrinsic parameter matrix and distortion coefficient of each camera are calculated using the detected corner points and actual physical coordinates.

[0084] Further, the light plane calibration method includes calculating the light plane by using the light plane equation, and calculating the light plane equation by fitting the point cloud;

[0085] The light plane equation is as follows:

[0086] ax+by+cz+d=0

[0087] Among them, (a, b, c) is the normal vector, and d is the distance from the plane to the origin;

[0088] Point cloud fitting involves fitting points on the light plane using the least squares method and calculating the light plane equation, which is as follows:

[0089] ax+by+cz+d=0

[0090] Among them, (a, b, c) is the normal vector and d is a constant term.

[0091] It should be noted that the specific steps of calibrating the plane in the camera coordinate system also include:

[0092] Laser installation: Fix the laser in a stable position and ensure that the emission plane is parallel to the calibration scene;

[0093] Set up the calibration board: Use the same checkerboard calibration board as the binocular calibration board, and ensure that the light plane has a clear light spot on the calibration board;

[0094] Capture images: The left and right cameras simultaneously capture images of multiple lasers irradiating the checkerboard calibration plate;

[0095] Extracting light spots: Detecting laser spots in images through image processing algorithms (threshold segmentation and morphological operations);

[0096] Spot center detection: calculate the centroid position of the spot as a point on the light plane;

[0097] Point cloud fitting: Fit the points on the light plane using the least squares method and calculate the light plane equation;

[0098] Coordinate transformation: transform the fitted light plane equation into the left and right camera coordinate systems to ensure accurate positioning in their respective coordinate systems.

[0099] Furthermore, the image processing includes image preprocessing and extraction of the laser line centerline by Steger algorithm;

[0100] Image preprocessing methods include iterative thresholding and binarization, as well as ROI template matching;

[0101] like Figure 2 As shown in (e), the thresholding method and binarization include preliminary selection of the threshold and iterative adjustment of the threshold until the image segmentation effect is satisfactory; the final threshold is applied to binarize the image. The calculation formula of the thresholding method and binarization is:

[0102]

[0103] Where T is the threshold, I filtered (x, y) is the image after Gaussian filtering, I binary (x, y) is a binary image;

[0104] like Figure 2 As shown in (f) and (g), the ROI template matching method includes further extracting the laser line area through template matching, removing redundant interference information, and performing template matching through sliding window and correlation calculation:

[0105]

[0106] Wherein, T is the template image, ω and h are the width and height of the template, respectively. The specific steps of the ROI template matching method are: defining an ROI template, which contains the expected laser line features;

[0107] Find the area that best matches the ROI template in the preprocessed image;

[0108] Based on the matching results, the exact position and shape of the laser line are further confirmed;

[0109] like Figure 2 As shown in (h), the Steger algorithm calculates the centerline intensity of each contour point and selects the position with the highest intensity as the corresponding point of the centerline, thereby forming the centerline of the laser line. The calculation formula is as follows:

[0110]

[0111] Among them, C(x, y) is the centerline intensity at point (x, y), I(x+u, y+v) is the image pixel value, G(u, v) is the Gaussian function used to weight the surrounding pixels, w is the normalization factor, and k determines the size of the window.

[0112] It should be noted that the image preprocessing also includes the following steps:

[0113] First, if Figure 2 As shown in (b), the original image is enhanced by the weighted grayscale method, that is, Figure 2 The brightness of the laser line in (a) is enhanced by contrast, that is, appropriate α and β values ​​are determined to enhance the brightness and contrast of the laser line, and the above formula is applied to each pixel in the image for weighted processing. The original image is grayed using the weighted formula:

[0114] I gray =0.2989×R+0.5870×G+0.1140×B

[0115] Apply weighting factors to adjust the contrast of a grayscale image:

[0116] I weighted =α×I gray +β

[0117] Among them, α is the contrast adjustment coefficient, and β is the brightness adjustment coefficient.

[0118] Secondly, if Figure 2 As shown in (c), through morphological processing, the expansion and corrosion operations in morphology are used to remove the noise in the image and retain the structural characteristics of the laser line. That is, the image is first opened (erode first and then expand) to remove small noise, and then closed (expand first and then erode) is performed to fill small holes.

[0119] Expansion operation:

[0120] Corrosion Operation:

[0121] Among them, SE is the structural element.

[0122] Finally, if Figure 2 As shown in (d), a Gaussian filter is used to process the image into a smooth image to reduce the influence of noise on laser line extraction, that is, a Gaussian filter G is designed to perform a convolution operation on the image to smooth the image and reduce the influence of noise. The formula for smoothing using a Gaussian filter is as follows:

[0123] Among them, G(i, j) is the convolution kernel of the Gaussian filter, and k is the radius of the kernel.

[0124] Furthermore, point cloud data acquisition includes obtaining the three-dimensional coordinates (X, Y, Z) of the point cloud data according to the pixel points (u, v) and depth information Z of the camera image coordinate system through the following formula:

[0125]

[0126] Among them, (c x , c y ) is the optical center coordinate, f x , f y is the focal length;

[0127] Point cloud data registration includes registering and fusing the point cloud contours of the left and right cameras through the ICP algorithm. The calculation steps include solving the rigid body transformation R and t that minimizes the mean square error between the source point cloud and the target point cloud. The formula is:

[0128]

[0129] By taking the derivative of the above formula and setting it to zero, we can get the following system of equations:

[0130]

[0131] Among them, R is the rotation matrix, t is the translation vector, and are the centroids of P and Q respectively, and U and V come from the covariance matrix The singular value decomposition of H=U∑V T , take R = I (unit matrix), t = 0 (zero vector);

[0132] The calculation step also includes, for the fused point cloud, using the local average method to reduce noise and smooth the point cloud contour, including calculating the average value of all noise points in the selected field point, that is, for each point P in the point cloud i (x i ,y i , z i ), find all the neighborhood points P within its neighborhood radius j (x j ,y j , z j ) is a point that satisfies the following conditions:

[0133]

[0134] The average value is calculated as follows:

[0135]

[0136] Where N is the number of neighborhood points, is the smoothed coordinate value.

[0137] It should be noted that a point cloud is a collection of many points, each of which has a coordinate value (x, y, z) in three-dimensional space; registration refers to aligning two or more point clouds from different sources so that they are represented in the same coordinate system. ICP is an iterative optimization algorithm used to align two point clouds. It achieves registration by repeatedly adjusting the position and posture of one point cloud to align it with the corresponding point of another point cloud. The goal of the ICP algorithm is to find a rigid body transformation, including a rotation matrix R and a translation vector t, so that the source point cloud P = p1, p2, ..., p n With the daily point cloud Q = q1, q2, ..., q n The specific steps are as follows:

[0138] Assume that the initial transformation is R = I (identity matrix) and t = 0 (zero vector);

[0139] For each point, find the nearest neighbor in the point cloud;

[0140] Solve the above formula to find the rigid body transformation R and t that minimizes the mean square error between the source point cloud and the target point cloud;

[0141] Update the position of the source point cloud using the calculated R and t.

[0142] Repeat the above steps until the error converges or the predetermined number of iterations is reached.

[0143] The streamlined point cloud contour is ICP-registered with the standard contour, including coarse registration and fine registration. Coarse registration reduces the error range through initial alignment, and fine registration further optimizes the alignment results to ensure high-precision matching between the point cloud data and the standard contour.

[0144] Rough registration: Use the initial transformation matrix T init , preprocess the simplified point cloud contour to make it as consistent as possible with the standard contour in position and direction. Suppose the simplified point set is P i , the rough registration calculation formula is as follows:

[0145] P′ i =T init ·P i

[0146] Among them, P i is the point cloud coordinate after initial transformation, T init is the initial transformation matrix containing rotation and translation.

[0147] Fine registration: Based on the rough registration, further optimize the alignment between the point cloud and the standard contour. The specific steps are as follows:

[0148] For each point in the transformed point cloud, find the nearest point in the standard contour as the corresponding point;

[0149] Calculate the new transformation matrix by minimizing the sum of squared distances between corresponding point pairs;

[0150] For the corresponding point pairs, calculate the new transformation matrix and apply the new transformation matrix to update the point cloud position. The calculation formula is as follows:

[0151]

[0152] Among them, |·| represents the Euclidean distance, and the updated point cloud position P″ i :P″ i =T ICP ;

[0153] The algorithm iterates until the transformation matrix T ICP The change in is less than the preset threshold or the maximum number of iterations is reached.

[0154] Furthermore, the rail wear calculation method includes performing polyline fitting according to the point cloud data to generate polyline functions f(x1) and f(x2) of the rail head, and obtaining the rail wear value by calculating the difference between the two functions at the vertical wear position and the side wear position, wherein the function form of the polyline fitting is:

[0155]

[0156] Among them, a i and b i are the slope and intercept of the fitted line segment, is the interval range of each line segment.

[0157] It should be noted that the method for calculating the wear value is that the vertical wear is the difference between the vertical direction of the rail to be tested and the standard section at one-third of the rail head width from the working edge. The horizontal wear is the difference between the horizontal direction of the rail to be tested and the standard section at 16 mm below the top of the rail. The total wear value is the vertical wear value plus half of the horizontal wear value.

[0158] In summary, the present invention registers the point clouds of the left and right cameras through the ICP algorithm to obtain a unified point cloud data set, which reflects the three-dimensional form of the same object observed from different perspectives. The fused point cloud allows the point clouds captured by the left and right cameras to be accurately aligned, eliminating the misalignment caused by different camera positions and angles, and the fused point cloud contains more detailed information because it integrates data from multiple perspectives. The registered point cloud can cover more parts of the object and improve the integrity and accuracy of the three-dimensional model. ICP also reduces errors through initial alignment of coarse registration, and then through the optimization step of fine registration, ensures high-precision matching between the simplified point cloud contour and the standard contour. This precise alignment is of great significance for subsequent analysis and application.

[0159] Example 2

[0160] Reference Fig. 9 , which is the second embodiment of the present invention, provides a rail wear measuring device, which can accurately shoot the rails from two perspectives to reduce the point cloud computing error.

[0161] Specifically, the carrying unit 100 includes a carrying assembly 101 and a supporting assembly 102 disposed at the lower side of the carrying assembly 101;

[0162] The measuring unit 200 includes a measuring component 201 disposed on the upper side of the carrier component 101 .

[0163] Furthermore, the object-carrying assembly 101 includes an object-carrying platform 101 a , and an installation groove 101 b is provided inside the object-carrying platform 101 a .

[0164] Furthermore, the support assembly 102 includes a fixed bracket 102a arranged on the lower side of the loading platform 101a, and a fixed foot 102b arranged on the outside of the fixed bracket 102a, wherein the fixed bracket 102a can adjust the height as needed, and the fixed foot 102b can be folded to the outside of the fixed bracket 102a when not placed.

[0165] Furthermore, the measuring component 201 includes a line laser 201a disposed on the upper side of the object carrying platform 101a, two industrial cameras 201b disposed on both sides of the line laser 201a, and a computer 201c connected to the industrial cameras 201b.

[0166] It should be noted that the industrial camera 201b and the line laser 201a are both installed in the mounting groove 101b, and can slide along the mounting groove 101b, which is convenient for multi-angle shooting of the rails. When shooting, they are fixedly connected to the loading platform 101a by bolts to prevent shaking during shooting and unclear shooting.

[0167] When in use, the staff opens the fixed foot 102b, adjusts the height of the fixed frame according to the position of the rail, and when it is adjusted to the appropriate height, fixes the fixed frame. Further, the staff installs the line laser 201a and the industrial camera 201b, adjusts the industrial camera 201b along the mounting groove 101b according to the required shooting position, and adjusts the line laser 201a as needed. When the adjustment of the industrial camera 201b and the line laser 201a is completed, the industrial camera 201b and the line laser 201a are fixed by bolts to complete the debugging before measurement.

[0168] When debugging is completed, turn on the line laser 201a to irradiate the rail with the line laser, and further control the industrial camera 201b through the computer 201c to shoot the rail and the line laser, and perform image processing and calculation through the computer 201c to finally obtain the rail wear value.

[0169] In summary, by adjusting the positions of the industrial camera 201b and the line laser 201a through the cooperation of the loading platform 101a and the fixed bracket 102a, the industrial camera 201b and the line laser 201a can be adaptively adjusted to the required positions while still being able to remain on the same horizontal line, ensuring that there is no excessive error in the alignment of the images taken by the two industrial cameras 201b. Through the cooperation of the two industrial cameras 201b and the line laser 201a, the position of the rail to be tested can be accurately photographed from a dual perspective, avoiding large errors during single-perspective shooting.

[0170] Importantly, it should be noted that the construction and arrangement of the present application shown in a plurality of different exemplary embodiments are only exemplary. Although only a few embodiments are described in detail in this disclosure, it should be readily understood by those who refer to this disclosure that many modifications are possible without substantially departing from the novel teachings and advantages of the subject matter described in the application (e.g., mounting arrangements, use of materials, color, changes in orientation, etc.). For example, the element shown as integrally formed may be composed of a plurality of parts or elements, the position of the element may be inverted or otherwise changed, and the nature or number or position of the discrete element may be altered or changed. Therefore, all such modifications are intended to be included within the scope of the present invention. The order or sequence of any process or method steps may be changed or reordered according to an alternative embodiment. In the claims, any "bracket plus function" clause is intended to cover the structure of the execution function described herein, and is not only structurally equivalent but also equivalent structure. Without departing from the scope of the present invention, other replacements, modifications, changes and omissions may be made in the design, operating conditions and arrangement of the exemplary embodiments. Therefore, the present invention is not limited to a specific embodiment, but extends to a variety of modifications that still fall within the scope of the appended claims.

[0171] Additionally, in an effort to provide a concise description of example embodiments, all features of an actual implementation may not be described.

[0172] It should be understood that in the development of any actual implementation, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but for those of ordinary skill having the benefit of this disclosure, the development effort will be a routine task of design, fabrication, and production without undue experimentation.

[0173] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A rail wear measurement method, characterized in that: include, Use the checkerboard method to perform binocular positioning and obtain the internal and external parameters of the left and right cameras; The light plane equation of the light plane in the left and right camera coordinate systems is obtained through light plane calibration; Image processing: the left and right cameras are controlled by a computer to collect multiple sets of images including the laser line profile containing rail information, and the original images are subjected to distortion correction processing; Acquire point cloud data and register the point cloud data; Calculate rail wear value.

2. The rail wear measurement method according to claim 1, characterized in that: The checkerboard method is a checkerboard pattern consisting of black and white squares, usually 9x6 or 12x9; The internal and external parameters of the left and right cameras include the internal parameter matrix and the external parameter matrix of the cameras on both sides. The expressions of the internal parameter matrix and the external parameter matrix are respectively expressed by the calibrateCamera function and the stereoCalibrate function. The expressions are as follows: Among them, f x 、f y is the focal length, c x 、c y is the principal point coordinate, R is the rotation matrix between cameras, T is the translation vector, and r is the distortion coefficient.

3. The rail wear measurement method according to claim 2, characterized in that: The light plane calibration method comprises calculating the light plane by using a light plane equation, and calculating the light plane equation by fitting a point cloud; The light plane equation is as follows: ax+by+cz+d=0 Among them, (a, b, c) is the normal vector, and d is the distance from the plane to the origin; The point cloud fitting includes fitting points on the light plane by the least square method and calculating the light plane equation, and the formula is as follows: ax+by+cz+d=0 Among them, (a, b, c) is the normal vector and d is a constant term.

4. The rail wear measurement method according to claim 3, characterized in that: The image processing includes image preprocessing and extracting the center line of the laser line by Steger algorithm; The image preprocessing method includes iterative threshold method and binarization, and ROI template matching method; The threshold generation method and binarization include initially selecting a threshold and iteratively adjusting the threshold until the image segmentation effect is satisfactory; Apply the final threshold to binarize the image. The calculation formula of the threshold method and binarization is: Where T is the threshold, I filtered (x, y) is the image after Gaussian filtering, I binary (x,y) is a binary image; The ROI template matching method includes further extracting the laser line area through template matching, removing redundant interference information, and performing template matching through sliding windows and correlation calculation: Where T is the template image, ω and h are the width and height of the template respectively; The Steger algorithm calculates the centerline intensity of each contour point and selects the position with the highest intensity as the corresponding point of the centerline, thereby forming the centerline of the laser line. The calculation formula is as follows: Among them, C(x,y) is the centerline intensity at point (x,y), I(x+u,y+v) is the image pixel value, G(u,v) is the Gaussian function used to weight the surrounding pixels, w is the normalization factor, and k determines the size of the window.

5. The rail wear measurement method according to claim 4, characterized in that: The point cloud data acquisition includes obtaining the three-dimensional coordinates (X, Y, Z) of the point cloud data according to the pixel points (u, v) and depth information Z of the camera image coordinate system through the following formula: Among them, (c x ,c y ) is the optical center coordinate, f x , f y is the focal length; The point cloud data registration includes registering and fusing the point cloud contours of the left and right cameras through the ICP algorithm. The calculation step includes solving the rigid body transformation R and t that minimizes the mean square error between the source point cloud and the target point cloud. The formula is: By taking the derivative of the above formula and setting it to zero, we can get the following system of equations: Among them, R is the rotation matrix, t is the translation vector, and are the centroids of P and Q respectively, and U and V come from the covariance matrix The singular value decomposition of H=UΣV T , take R = I (unit matrix), t = 0 (zero vector); The calculation step also includes, for the fused point cloud, using a local average method to perform noise reduction and smoothing on the point cloud contour, including calculating the average value of all noise points in the selected field point, that is, for each point P in the point cloud i (x i ,y i ,z i ), find all the neighboring points R within its neighborhood radius j (x j ,y j ,z j ) is a point that satisfies the following conditions: The average value is calculated as follows: Where N is the number of neighborhood points, is the smoothed coordinate value.

6. The rail wear measurement method according to claim 5, characterized in that: The rail wear calculation method includes performing polyline fitting according to point cloud data to generate polyline functions f(x1) and f(x2) of the rail head, and obtaining the rail wear value by calculating the difference between the two functions at the vertical wear position and the side wear position, wherein the function form of the polyline fitting is: Among them, a i and b i are the slope and intercept of the fitted line segment, is the interval range of each line segment.

7. A rail wear measuring device, characterized in that: include, A carrying unit (100) comprises a carrying assembly (101) and a supporting assembly (102) arranged at the lower side of the carrying assembly (101); The measuring unit (200) comprises a measuring component (201) arranged on the upper side of the object carrying component (101).

8. The rail wear measuring device according to claim 7, characterized in that: The object carrying assembly (101) comprises an object carrying platform (101a), and a mounting groove (101b) is provided on the inner side of the object carrying platform (101a).

9. The rail wear measuring device according to claim 8, characterized in that: The support assembly (102) comprises a fixed bracket arranged on the lower side of the loading platform (101a), and a fixed foot (102b) arranged on the outer side of the fixed bracket (102a).

10. The rail wear measuring device according to claim 9, characterized in that: The measuring component (201) comprises a line laser (201a) arranged on the upper side of the object carrying platform (101a), two industrial cameras (201b) arranged on both sides of the line laser (201a), and a computer (201c) connected to the industrial cameras (201b).

Citation Information

Cited By

  • Non-contact rail abrasion and third rail geometric parameter real-time detection method and system

    CN119714066A

  • Jump gap rapid measurement method and device based on visual guidance

    CN121207058A