Machine vision-based method for measuring distance between cross positions of carrier cables of rail contact network

Abstract

The invention discloses a method and a system for measuring the distance between cross positions of carrier cables of a rail contact network based on machine vision, and relates to the technical field of measurement of the distance between the cross positions of the carrier cables of the rail contact network. A laser radar and an infrared sensor are used for collecting depth information and temperature distribution data, and a multi-view image and depth temperature data are obtained; collecting the multi-view image and the depth temperature data to a central processing unit through a high-speed data transmission channel to obtain an initial data set; preprocessing the initial data set to obtain a high-quality data set; based on a deep learning algorithm and the high-quality data set, constructing a prediction model of the carrier cable and the intersection point, inputting sample data in the high-quality data set, and outputting prediction results of the carrier cable and the intersection point; and based on prediction results of the carrier cables and the intersection points, three-dimensional model reconstruction is carried out on the intersection positions of the carrier cables in combination with stereoscopic vision and structured light technologies.

Description

Technical Field

[0001] The present invention relates to the technical field of measuring the spacing between the cross positions of the catenary suspension cables of an overhead contact line, and particularly to a method for measuring the spacing between the cross positions of the catenary suspension cables of an overhead contact line based on machine vision. Background Art

[0002] Measuring the spacing between the cross positions of the catenary suspension cables of an overhead contact line refers to the process of accurately measuring the relative distance between the catenary suspension cables used to hang the contact wire in rail transit systems such as railways or subways. Therefore, how to use advanced technical means to improve the intelligent level and safety of measuring the spacing between the cross positions of the catenary suspension cables of an overhead contact line has become one of the urgent problems to be solved currently.

[0003] In the technical field of measuring the spacing between the cross positions of the catenary suspension cables of an overhead contact line, traditional methods only rely on a single type of sensor (such as a camera) for data acquisition, and cannot simultaneously obtain multi-dimensional data such as depth information and temperature distribution. Moreover, traditional 3D reconstruction methods cannot accurately capture complex geometric shapes and detailed features, resulting in inaccurate reconstruction models. At the same time, traditional methods require a large amount of manual intervention for image annotation, feature extraction, etc., which is not only time-consuming and laborious but also prone to errors. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for measuring the spacing between the cross positions of the catenary suspension cables of an overhead contact line based on machine vision to solve the problems that traditional methods only rely on a single type of sensor (such as a camera) for data acquisition, cannot simultaneously obtain multi-dimensional data such as depth information and temperature distribution, and traditional 3D reconstruction methods cannot accurately capture complex geometric shapes and detailed features.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for measuring the spacing between the cross positions of the catenary suspension cables of an overhead contact line based on machine vision, which includes:

[0008] Taking pictures of the overhead contact line area from multiple angles using a high-resolution camera, and collecting depth information and temperature distribution data using a lidar and an infrared sensor to obtain multi-view images and depth-temperature data;

[0009] Collecting the multi-view images and depth-temperature data to a central processing unit through a high-speed data transmission channel to obtain an initial data set;

[0010] Preprocessing the initial data set to obtain a high-quality data set;

[0011] Build a prediction model for the catenary and intersection points based on deep learning algorithms and high-quality datasets, input the sample data in the high-quality dataset, and output the prediction results of the catenary and intersection points;

[0012] Based on the prediction results of the catenary and intersection points, combined with stereo vision and structured light technology, reconstruct the 3D model of the catenary intersection position, and obtain the reconstructed 3D model;

[0013] Calculate the exact distance between the catenary intersection positions based on the spatial coordinates in the 3D model to obtain the distance measurement result;

[0014] Analyze the distance measurement result to determine whether there are potential safety hazards, and generate a detailed inspection report to obtain an inspection report including safety analysis.

[0015] As a preferred solution of the method for measuring the distance between the catenary intersection positions of the overhead catenary based on machine vision according to the present invention, wherein: using a high-resolution camera to take pictures of the overhead catenary area from multiple angles, and using a lidar and an infrared sensor to collect depth information and temperature distribution data to obtain multi-view images and depth-temperature data, the specific steps are as follows:

[0016] Prepare a high-resolution camera Sony IMX, and configure a lidar Velodyne VLP-16 and an infrared sensor FLIR thermal imager;

[0017] Install at least three high-resolution cameras above the overhead catenary, and synchronously start all sensors, including high-resolution cameras, lidars, and infrared sensors, to start data collection;

[0018] Each camera takes a set of photos, and at the same time records the point cloud data obtained by the lidar scan and the temperature distribution map measured by the infrared sensor at the corresponding moment;

[0019] Use a standard target board to geometrically calibrate the camera to ensure that the images from different perspectives can be accurately aligned to obtain multi-view images and depth-temperature data.

[0020] As a preferred solution of the method for measuring the distance between the catenary intersection positions of the overhead catenary based on machine vision according to the present invention, wherein: collecting the multi-view images and depth-temperature data to the central processing unit through a high-speed data transmission channel to obtain an initial dataset, the specific steps are as follows:

[0021] Use a high-speed data transmission interface GigE Vision to connect the sensor to the central processing unit and configure the network parameters;

[0022] Associate the preliminarily sorted different types of data according to the time sequence and spatial position relationship to obtain an initial dataset.

[0023] As a preferred solution of the method for measuring the spacing of the catenary cross positions based on machine vision according to the present invention, wherein: the preprocessing of the initial data set to obtain a high-quality data set is specifically carried out as follows:

[0024] Apply a Gaussian filter to smooth the multi-view images in the initial data set to remove random noise in the images;

[0025] Apply bilateral filtering to the point cloud data collected by the lidar to reduce noise interference while retaining edge information;

[0026] Use the known internal and external parameters of the camera to geometrically calibrate the multi-view images, detect and match feature points using the SIFT algorithm, and calculate the transformation matrix through the RANSAC algorithm to achieve image registration;

[0027] Aiming at the color deviation problem of multi-view images, use the white balance algorithm to adjust the colors of each image to make their colors consistent;

[0028] Fuse the multi-source data processed above into a unified space model, optimize the point cloud registration accuracy using the ICP algorithm, adjust it in combination with the temperature distribution information, and obtain a high-quality data set.

[0029] As a preferred solution of the method for measuring the spacing of the catenary cross positions based on machine vision according to the present invention, wherein: the prediction model of the catenary and the cross points is constructed based on the deep learning algorithm and the high-quality data set, the sample data in the high-quality data set is input, and the prediction results of the catenary and the cross points are output. The specific steps are as follows:

[0030] Annotate the images and three-dimensional space data in the high-quality data set to mark the exact positions of all catenaries and their cross points;

[0031] Divide the annotated high-quality data set into a training set, a validation set, and a test set;

[0032] Based on the deep learning algorithm CNN and the annotated high-quality data set, construct a prediction model of the catenary and the cross points.

[0033] Substitute the high-quality data set into the prediction model of the catenary and the cross points, and output the prediction results of the catenary and the cross points.

[0034] As a preferred solution of the method for measuring the spacing of the catenary cross positions based on machine vision according to the present invention, wherein: based on the prediction results of the catenary and the cross points, combined with stereo vision and structured light technology, three-dimensional model reconstruction of the catenary cross position is carried out, and the reconstructed three-dimensional model is obtained. The specific steps are as follows:

[0035] Calculate the disparity map through a stereo vision algorithm using high-resolution images from multiple perspectives;

[0036] Use a structured light device to project light with a known pattern onto the overhead catenary area and capture the reflected pattern;

[0037] Calculate the three-dimensional coordinates of the object surface based on the deformation of the light;

[0038] Combine the depth map, the three-dimensional point cloud data obtained through structured light technology, and the position information of the catenary and its intersections extracted from the prediction result R of the catenary and intersections;

[0039] Use the ICP algorithm for data fusion and reconstruction to obtain the reconstructed three-dimensional model.

[0040] As a preferred solution of the method for measuring the distance between catenary intersections of an overhead catenary based on machine vision according to the present invention, wherein: calculating the exact distance between catenary intersections based on the spatial coordinates in the three-dimensional model to obtain a distance measurement result, and the specific steps are as follows:

[0041] Extract the spatial coordinates of the catenary intersections in the reconstructed three-dimensional model M;

[0042] For any two intersections, calculate the Euclidean distance between them according to the spatial coordinates;

[0043] Output the distance measurement results for each pair of catenary intersection positions.

[0044] As a preferred solution of the method for measuring the distance between catenary intersections of an overhead catenary based on machine vision according to the present invention, wherein: analyzing the distance measurement results to determine whether there are potential safety hazards and generating a detailed inspection report to obtain an inspection report including safety analysis, and the specific steps are as follows:

[0045] Collect all the exact distance values calculated between catenary intersections;

[0046] According to safety standards, set safety thresholds for the distance between each type of catenary intersection;

[0047] Compare each measured distance value with the corresponding safety threshold to determine whether there are potential safety hazards;

[0048] Based on the safety assessment result A, conduct a detailed safety analysis for each detection point and obtain an inspection report including safety analysis.

[0049] In a second aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the method for measuring the distance between the catenary cross positions based on machine vision as described in the first aspect of the present invention is implemented.

[0050] In a third aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the method for measuring the distance between the catenary cross positions based on machine vision as described in the first aspect of the present invention is implemented.

[0051] The beneficial effects of the present invention are as follows: By using a high-resolution camera to capture images of the overhead catenary area from multiple angles, and using lidar and infrared sensors to collect depth information and temperature distribution data, a comprehensive perception of the overhead catenary environment is achieved. By preprocessing the initial data set, a high-quality data set is obtained, significantly reducing the interference factors in the data, improving the availability and consistency of the data, enhancing the accuracy and reliability of the data, and providing high-quality input for the subsequent deep learning model. Based on the deep learning algorithm and the high-quality data set, a prediction model for the catenary and intersection points is constructed, and the sample data in the high-quality data set is input, and the prediction results of the catenary and intersection points are output. By using advanced deep learning technology to automatically identify and locate the positions of the catenary and its intersection points, the accuracy and efficiency of the identification are greatly improved. While reducing manual intervention, the accuracy of the position identification is ensured, achieving the effects of high automation and high identification accuracy, and greatly improving the work efficiency and accuracy. Description of the Drawings

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0053] Figure 1 It is a flowchart of the method for measuring the distance between the catenary cross positions based on machine vision in Embodiment 1. Detailed Embodiments

[0054] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention with reference to the drawings in the specification.

[0055] 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.

[0056] 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.

[0057] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides a method for measuring the crossing position spacing of catenary cables of a track contact network based on machine vision, comprising the following steps:

[0058] S1. Use high-resolution cameras to photograph the track contact network area from multiple angles, and use laser radar and infrared sensors to collect depth information and temperature distribution data to obtain multi-view images and depth temperature data;

[0059] Going a step further, prepare a high-resolution camera Sony IMX, and configure the lidar Velodyne VLP-16 and infrared sensor FLIR thermal imager;

[0060] Install at least three high-resolution cameras above the track contact network and start all sensors, including high-resolution cameras, lidar and infrared sensors, to collect data simultaneously;

[0061] Each camera takes a set of photos and records the point cloud data obtained by the lidar scan at the corresponding moment and the temperature distribution map measured by the infrared sensor;

[0062] Use a standard target plate to perform geometric calibration on the camera to ensure that images at different viewing angles can be accurately aligned to obtain multi-view images and depth temperature data;

[0063] It should be noted that the combination of high-resolution camera Sony IMX, lidar Velodyne VLP-16 and infrared sensor FLIR thermal imager can not only provide high-quality visual images, but also obtain accurate depth information and temperature distribution data. This multi-sensor fusion method ensures the comprehensiveness and accuracy of the data, providing a solid foundation for subsequent data processing and analysis.

[0064] S2, collecting the multi-view images and depth temperature data to the central processing unit through a high-speed data transmission channel to obtain an initial data set;

[0065] Furthermore, connect the sensor to the central processing unit using the high-speed data transmission interface GigE Vision and configure the network parameters;

[0066] Associate the preliminarily sorted different types of data according to the time sequence and spatial position relationship to obtain an initial data set;

[0067] It should be noted that connecting the sensor to the central processing unit using the high-speed data transmission interface GigE Vision and configuring the network parameters can ensure that data will not be lost or delayed during transmission, thus guaranteeing the consistency and integrity of the data. Associating different types of data according to the time sequence and spatial position relationship helps to improve the efficiency and accuracy of subsequent data processing and model training.

[0068] S3. Preprocess the initial data set to obtain a high-quality data set;

[0069] Furthermore, apply a Gaussian filter to smooth the multi-view images in the initial data set and remove random noise in the images;

[0070] Apply bilateral filtering to the point cloud data collected by the lidar to reduce noise interference while retaining edge information;

[0071] Use the known internal and external parameters of the camera to geometrically calibrate the multi-view images, and use the SIFT algorithm to detect and match feature points. Calculate the transformation matrix through the RANSAC algorithm to achieve image registration;

[0072] For the color deviation problem of multi-view images, use the white balance algorithm to adjust the colors of each image to make their colors consistent. The expression is:

[0073]

[0074] Among them, C(r, g, b) represents the value of the original image in the RGB color space, C′(r, g, b) represents the color value of the image after white balance correction, and W r ,W g ,W b are the weight coefficients of the red, green, and blue channels respectively;

[0075] Fuse the multi-source data processed above into a unified space model, use the ICP algorithm to optimize the point cloud registration accuracy, and adjust it in combination with the temperature distribution information to obtain a high-quality data set;

[0076] It should be noted that through preprocessing steps such as applying Gaussian filters and smoothing to remove image noise, bilateral filtering to optimize point cloud data, geometric calibration, and feature point matching, the quality and consistency of the data have been significantly improved. The processing not only reduces interference factors in the data but also ensures that images from different perspectives can be accurately aligned, providing high-quality input for subsequent 3D reconstruction and deep learning models.

[0077] S4. Based on deep learning algorithms and high-quality datasets, construct a prediction model for catenaries and intersection points, input the sample data in the high-quality dataset, and output the prediction results of catenaries and intersection points;

[0078] Furthermore, annotate the images and 3D spatial data in the high-quality dataset to mark the exact positions of all catenaries and their intersection points;

[0079] Divide the annotated high-quality dataset into a training set, a validation set, and a test set;

[0080] Based on the deep learning algorithm CNN and the annotated high-quality dataset, construct a prediction model for catenaries and intersection points, and the expression is:

[0081] M = CNN(I, D);

[0082] Where M is the prediction model for catenaries and intersection points, I is the high-quality dataset, D is the point cloud data in 3D space, and CNN(.) is a composite function.

[0083] Substitute the high-quality dataset into the prediction model for catenaries and intersection points, and output the prediction results of catenaries and intersection points. The expression is:

[0084] R = M(I);

[0085] Where R is the prediction result of catenaries and intersection points, M is the prediction model for catenaries and intersection points, and I is the high-quality dataset;

[0086] It should be noted that by constructing a prediction model for catenaries and intersection points based on the deep learning algorithm CNN and training with the annotated high-quality dataset, automated recognition and positioning can be achieved. This method not only improves the recognition accuracy and efficiency but also reduces the need for manual intervention, making the entire process more efficient and reliable. The high-quality dataset is the key guarantee for the model's performance, ensuring the accuracy and reliability of the prediction results.

[0087] S5. Based on the prediction results of catenaries and intersection points, combine stereo vision and structured light technology to perform 3D model reconstruction on the catenary intersection positions and obtain the reconstructed 3D model;

[0088] Further, using high-resolution images from multiple perspectives, a disparity map is calculated through a stereo vision algorithm, with the expression:

[0089] D = S(I1, I2);

[0090] Among them, D represents the generated depth map, I1 and I2 respectively represent two images from different perspectives, and S(.) represents the stereo matching algorithm, which is used to calculate the disparity and generate the depth map;

[0091] Use a structured light device to project light rays of a known pattern onto the overhead catenary area and capture the reflected pattern;

[0092] Calculate the three-dimensional coordinates of the object surface according to the deformation of the light rays, with the expression:

[0093] P = L(C, Pp);

[0094] Among them, P represents the three-dimensional point cloud data obtained through the structured light technology, C represents the captured light ray deformation pattern, Pp represents the projected known light ray pattern, and L(.) represents the structured light algorithm, which is used to calculate the three-dimensional coordinates;

[0095] Combine the depth map, the three-dimensional point cloud data obtained through the structured light technology, and the position information of the catenary and its intersections extracted from the prediction result R of the catenary and intersections;

[0096] Use the ICP algorithm for data fusion and reconstruction, and obtain the reconstructed three-dimensional model, with the expression:

[0097] M = F(D, P, R);

[0098] Among them, M represents the reconstructed three-dimensional model, and F(.) represents the fusion algorithm, which is used to integrate multi-source data and generate the three-dimensional model;

[0099] It should be noted that combining the stereo vision and the structured light technology for three-dimensional model reconstruction can not only generate an accurate three-dimensional model, but also capture complex geometric shapes and detailed features. The method makes full use of the advantages of multiple sensors to ensure the integrity and accuracy of the reconstructed model. The application of the ICP algorithm further optimizes the point cloud registration accuracy and improves the quality of the three-dimensional model.

[0100] S6. Calculate the exact distance between the catenary intersection positions based on the spatial coordinates in the three-dimensional model to obtain the distance measurement result;

[0101] Further, extract the spatial coordinates of the catenary intersections in the reconstructed three-dimensional model M;

[0102] For any two intersections, calculate the Euclidean distance between them according to the spatial coordinates;

[0103] Output the distance measurement results between each pair of catenary crossing positions;

[0104] It should be noted that calculating the exact distance between catenary crossing positions based on the spatial coordinates in the 3D model can provide very accurate distance measurement results. This step is crucial for evaluating the safety status of the overhead contact line of the track because the exact distance information can directly reflect the safety of the structure, avoid potential risks, and the positional relationship of all key points can be quickly and effectively obtained by calculating the Euclidean distance between any two crossing points.

[0105] S7. Analyze the distance measurement results to determine whether there are potential safety hazards and generate a detailed inspection report to obtain an inspection report containing safety analysis;

[0106] Furthermore, collect all the exact distance values between the calculated catenary crossing positions;

[0107] According to the safety standards, set the safety thresholds for the spacing of each type of catenary crossing position;

[0108] Compare each measured distance value with the corresponding safety threshold to determine whether there are potential safety hazards. The expression is:

[0109] A = W(N, T);

[0110] Where, A represents the safety assessment result, N is the distance measurement result, T is the safety threshold, and W(.) represents the comparison and assessment process;

[0111] Based on the safety assessment result A, conduct a detailed safety analysis on each detection point and obtain an inspection report containing safety analysis;

[0112] It should be noted that conducting a safety assessment on the distance measurement results and generating a detailed inspection report can not only timely detect potential safety hazards but also provide a scientific basis for maintenance work. According to the thresholds set by the safety standards, the safety status of each measurement point can be quantified, and specific improvement suggestions can be put forward through a detailed safety analysis report. This method improves the pertinence and effectiveness of maintenance work, helps to extend the service life of facilities, and reduces the incidence of safety accidents.

[0113] This embodiment also provides a computer device applicable to the case of the method for measuring the spacing of catenary crossing positions of the overhead contact line of the track based on machine vision, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for measuring the spacing of catenary crossing positions of the overhead contact line of the track based on machine vision as proposed in the above embodiment.

[0114] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be achieved through Wi-Fi, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, trackball, or touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0115] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for measuring the distance between the cross positions of the catenary messenger wires based on machine vision as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM for short), Electrically Erasable Programmable Read-Only Memory (EEPROM for short), Erasable Programmable Read Only Memory (EPROM for short), Programmable Red-Only Memory (PROM for short), Read-Only Memory (ROM for short), magnetic memory, flash memory, a magnetic disk, or an optical disc.

[0116] In summary, the present invention comprehensively perceives the environment of the overhead catenary by using a high-resolution camera to capture the overhead catenary area from multiple angles and using lidar and infrared sensors to collect depth information and temperature distribution data. By preprocessing the initial dataset, a high-quality dataset is obtained, significantly reducing the interference factors in the data, improving the usability and consistency of the data, enhancing the accuracy and reliability of the data, and providing high-quality input for the subsequent deep learning model. Based on the deep learning algorithm and the high-quality dataset, a prediction model for the messenger wire and the crossover point is constructed, and the sample data in the high-quality dataset is input, and the prediction results of the messenger wire and the crossover point are output. The advanced deep learning technology is used to automatically identify and locate the position of the messenger wire and its crossover point, greatly improving the accuracy and efficiency of the identification. While reducing manual intervention, the accuracy of the position identification is ensured, achieving the effects of high automation and high identification accuracy, and greatly improving the work efficiency and accuracy.

[0117] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for measuring the cross position spacing of catenary cables of a track contact network based on machine vision, characterized in that: include: Use high-resolution cameras to photograph the track contact network area from multiple angles, and use lidar and infrared sensors to collect depth information and temperature distribution data to obtain multi-view images and depth temperature data; The multi-view images and depth temperature data are collected to the central processing unit through a high-speed data transmission channel to obtain an initial data set; Preprocess the initial data set to obtain a high-quality data set; Construct a prediction model for catenary cables and intersections based on deep learning algorithms and high-quality data sets, input sample data from high-quality data sets, and output prediction results for catenary cables and intersections; Based on the prediction results of the catenary cables and intersections, the three-dimensional model of the intersection position of the catenary cables is reconstructed by combining stereo vision and structured light technology, and the reconstructed three-dimensional model is obtained; Calculate the precise distance between the cross positions of the catenary cables based on the spatial coordinates in the three-dimensional model to obtain the distance measurement result; The distance measurement results are analyzed to determine whether there are any safety hazards, and a detailed test report is generated to obtain a test report including a safety analysis.

2. The method for measuring the crossing position spacing of catenary cables of a track contact network based on machine vision according to claim 1, characterized in that: The method uses a high-resolution camera to photograph the track contact network area from multiple angles, and uses a laser radar and an infrared sensor to collect depth information and temperature distribution data to obtain multi-view images and depth temperature data. The specific steps are as follows: Prepare a high-resolution camera Sony IMX, and configure a lidar Velodyne VLP-16 and an infrared sensor FLIR thermal imager; Install at least three high-resolution cameras above the track contact network and start all sensors, including high-resolution cameras, lidar and infrared sensors, to collect data simultaneously; Each camera takes a set of photos and records the point cloud data obtained by the lidar scan at the corresponding moment and the temperature distribution map measured by the infrared sensor; The camera is geometrically calibrated using a standard target plate to ensure that images at different viewing angles can be accurately aligned to obtain multi-view images and depth temperature data.

3. The method for measuring the crossing position spacing of catenary cables of a track contact network based on machine vision as claimed in claim 2, characterized in that: The multi-view images and depth temperature data are collected to the central processing unit through a high-speed data transmission channel to obtain an initial data set. The specific steps are as follows: Use the high-speed data transmission interface GigE Vision to connect the sensor to the central processing unit and configure the network parameters; The different types of data that have been preliminarily sorted are associated according to time sequence and spatial position relationship to obtain the initial data set.

4. The method for measuring the crossing position spacing of catenary cables of a track contact network based on machine vision as claimed in claim 3, characterized in that: The initial data set is preprocessed to obtain a high-quality data set, and the specific steps are as follows: Apply Gaussian filter to smooth the multi-view images in the initial data set to remove random noise in the images; Apply bilateral filtering to the point cloud data collected by the LiDAR to reduce noise interference while retaining edge information; The multi-view images are geometrically calibrated using known camera internal and external parameters, and the SIFT algorithm is used to detect and match feature points. The RANSAC algorithm is used to calculate the transformation matrix to achieve image registration. To solve the color deviation problem of multi-view images, a white balance algorithm is used to adjust the colors of each image to make them consistent. The multi-source data after the above processing are fused into a unified spatial model, and the ICP algorithm is used to optimize the point cloud registration accuracy, and the temperature distribution information is combined for adjustment to obtain a high-quality data set.

5. The method for measuring the crossing position spacing of catenary cables of a track contact network based on machine vision as claimed in claim 4, characterized in that: The prediction model of the catenary cables and intersections is constructed based on the deep learning algorithm and the high-quality data set, sample data in the high-quality data set is input, and the prediction results of the catenary cables and intersections are output. The specific steps are as follows: Annotate images and 3D spatial data in high-quality datasets to mark the exact locations of all catenary cables and their intersections; Divide the labeled high-quality dataset into training set, validation set and test set; Based on the deep learning algorithm CNN and annotated high-quality data sets, a prediction model for catenary cables and intersections is constructed; Substitute the high-quality data set into the prediction model of the catenary cables and intersections, and output the prediction results of the catenary cables and intersections.

6. The method for measuring the crossing position spacing of catenary cables of a track contact network based on machine vision as claimed in claim 5, characterized in that: Based on the prediction results of the catenary cables and the intersection points, the three-dimensional model of the intersection position of the catenary cables is reconstructed in combination with stereo vision and structured light technology, and the reconstructed three-dimensional model is obtained. The specific steps are as follows: Using high-resolution images from multiple perspectives, the disparity map is calculated using a stereo vision algorithm; Use structured light equipment to project a known pattern of light onto the track catenary area and capture the reflected pattern; Calculate the three-dimensional coordinates of the object surface based on the light deformation; Combining the depth map, the 3D point cloud data obtained by structured light technology, and the position information of the catenary cables and their intersections extracted from the prediction results R of the catenary cables and their intersections; The ICP algorithm is used for data fusion and reconstruction to obtain the reconstructed three-dimensional model.

7. The method for measuring the crossing position spacing of catenary cables of a track contact network based on machine vision as claimed in claim 6, characterized in that: The precise distance between the cross positions of the load-bearing cables is calculated based on the spatial coordinates in the three-dimensional model to obtain the distance measurement result. The specific steps are as follows: Extracting the spatial coordinates of the cross points of the load-bearing cables in the reconstructed three-dimensional model M; For any two intersection points, the Euclidean distance between them is calculated based on the spatial coordinates; Output the distance measurements between each pair of catenary cable crossing locations.

8. The method for measuring the crossing position spacing of catenary cables of a track contact network based on machine vision according to claim 7, characterized in that: The distance measurement results are analyzed to determine whether there are safety hazards, and a detailed test report is generated to obtain a test report including a safety analysis. The specific steps are: Collect the calculated exact distance values ​​between all catenary cable crossing positions; According to the safety standards, set the safety threshold of the spacing between the cross positions of each type of load-bearing cables; Compare each measured distance value with the corresponding safety threshold to determine whether there is a safety hazard; Based on the safety assessment result A, a detailed safety analysis is performed on each detection point, and a detection report including the safety analysis is obtained.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method for measuring the crossing position spacing of the catenary cables of the track contact network based on machine vision are implemented as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for measuring the crossing position spacing of the catenary cables of a track contact network based on machine vision according to any one of claims 1 to 7 are implemented.

Images