A computer vision-based intelligent detection method and system for railway platform clearance.

Abstract

This invention relates to the field of computer vision technology, specifically to a computer vision-based intelligent detection method and system for station platform clearances. The method includes the following steps: acquiring spectral data of the station area; extracting spectral features based on the reflectance value of each pixel within the spectral range and comparing them with features of various known materials to identify material types at multiple locations; detecting obstacles; and generating material classification feature values. In this invention, material types are identified by acquiring spectral data and comparing reflectance values; texture features and gradient change rate analysis are used to identify blurred areas, reducing misjudgments of blurred markings; spatiotemporal registration of point cloud data and image information is employed; multi-source perception information is fused to enhance the spatiotemporal consistency of 3D reconstruction, improving the dimensional accuracy of the station model; spatial offset is calculated by comparing with engineering design drawings; a quantitative evaluation system for structural deviations is established; deformed areas are located, improving the predictability of safety warnings.

Description

Technical Field

[0001] This invention relates to the field of computer vision technology, and in particular to a computer vision-based intelligent detection method and system for station platform clearance. Background Technology

[0002] The field of computer vision technology encompasses the analysis, understanding, and processing of image and video data using computers and algorithms. Its core components include image acquisition, feature extraction, object detection, object recognition, image segmentation, 3D reconstruction, and image enhancement. Computer vision technology is widely used in intelligent monitoring, autonomous driving, medical image analysis, industrial inspection, and facial recognition. Its development primarily relies on deep learning, traditional machine learning, and pattern recognition technologies. By constructing neural networks or rule-based methods, it achieves automatic analysis and understanding of image and video data. Combining data preprocessing, feature representation, classifier training, and object tracking, different optimization algorithms and computational models are typically employed for different application scenarios to improve recognition accuracy and computational efficiency.

[0003] Among them, the intelligent detection method for station platform clearance refers to the use of computer vision technology to automatically detect the clearance status of station platforms. Through the acquisition of image or video data, combined with feature extraction, target detection and 3D measurement, elements such as platform edges, obstacles and clearance signs are identified and located. This includes platform structure reconstruction based on multi-view images, target detection based on deep learning, and distance calculation based on point cloud data. It accurately identifies the platform area and detects its clearance status. Using image matching and feature point tracking technology, it realizes dynamic monitoring of platform facilities and surrounding environment, and performs data analysis on factors that may affect train operation safety.

[0004] Traditional station platform clearance detection technology relies on single visual feature analysis, lacking the ability to distinguish the material characteristics of the platform. When the reflective properties of the obstacle surface are similar to the background, it is prone to missed detections and misjudgments. It does not consider feature degradation caused by natural wear or light interference. The lack of a gradient direction interpolation mechanism limits the positioning accuracy of fuzzy marked areas. The spatiotemporal registration accuracy of point cloud and image data is insufficient. The lack of a time synchronization mechanism for sensor data leads to 3D model jitter or misalignment during dynamic monitoring, affecting the continuous tracking of structural deformation. Structural offset assessment lacks a benchmark for comparison with engineering design data. Relying solely on on-site measurement data makes it difficult to distinguish between construction errors and subsequent deformation, affecting the targeted nature of maintenance decisions. Collision risk assessment does not integrate train operating parameters. Static clearance detection is disconnected from dynamic operating trajectory, making it impossible to predict changes in safety margins under special conditions such as curve driving or emergency braking, resulting in a risk of delayed prediction of safety hazards. Summary of the Invention

[0005] To address the technical problems existing in the prior art, this invention provides a computer vision-based intelligent detection method and system for station platform clearance. The technical solution is as follows:

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a computer vision-based intelligent detection method for station platform clearance, comprising the following steps:

[0007] S1: Acquire spectral data of the platform area, extract spectral features based on the reflectance value of each pixel within the spectral range, compare them with the features of various known materials, identify the material types at multiple locations, detect obstacles, and generate material classification feature values.

[0008] S2: Based on the material classification feature value, use an image acquisition device to acquire an image of the platform area, extract the texture features of the boundary marker area, identify the blurred area of ​​the marker by calculating the directional gradient change rate, and calculate the gradient direction interpolation of the blurred area to obtain the boundary marker correction value.

[0009] S3: Based on the clearance mark correction value, use lidar to acquire point cloud data of the platform area, combine with image information, perform spatiotemporal synchronization registration of the platform point cloud data, and construct a platform structure model.

[0010] S4: Based on the platform structure model, by comparing it with the platform structure data in the engineering design drawings, calculate the spatial offset and offset angle of multiple points, and generate the structural offset value;

[0011] S5: Based on the structural offset value, extract the offset point information, call the train's outer dimensions and running trajectory, calculate the collision risk between the platform clearance and the train, and generate a collision risk value.

[0012] As a further aspect of the present invention, the material classification feature value specifically includes the material spectral reflectance parameter, the material type matching result, and the obstacle spectral feature value; the clearance marker correction value includes the clearance marker texture direction gradient change rate, the fuzzy region gradient interpolation calculation value, and the marker boundary correction parameter; the platform structure model specifically includes the platform edge point cloud coordinates, clearance marker point cloud data, and platform facility point cloud data; the structure offset value includes the change in spatial coordinates of the offset point, the change in the rotation angle of the offset point, and the platform structure offset trend data; and the collision risk value includes the platform offset distance, the platform offset angle, and the collision risk level.

[0013] As a further aspect of the present invention, the steps of acquiring spectral data of the platform area, extracting spectral features based on the reflectance value of each pixel within the spectral range, comparing them with the features of various known materials, identifying material types at multiple locations, detecting obstacles, and generating material classification feature values ​​are as follows:

[0014] S101: Acquire spectral data of the platform area, extract the reflectance value of each pixel, calculate the spectral peak position, bandwidth half width at half maximum, and spectral slope parameters of multiple bands, extract spectral features of multiple positions, and generate a set of spectral feature parameters.

[0015] S102: Based on the spectral feature parameter set, call the spectral database of known materials, calculate the spectral similarity of materials at multiple locations on the platform, calibrate the material type at multiple locations on the platform surface, and obtain the platform material category distribution data;

[0016] S103: Call the platform material category distribution data, detect spectral reflectance anomalies in the platform area, identify obstacles, record location information in real time, and generate material classification feature values.

[0017] As a further aspect of the present invention, the specific formula for calculating the material spectral similarity at multiple locations of the platform is as follows:

[0018]

[0019] Calculate the spectral matching index;

[0020] Where S′ represents the matching degree index between the platform's measured spectrum and the known material spectrum, A k B represents the reflectance value of the spectrum measured at the k-th wavelength point on the platform. k This represents the reflectance value of the known material spectrum at the k-th wavelength point, where K represents the total number of wavelength points used to calculate the spectral matching degree, and k represents the wavelength point index in the spectral measurement.

[0021] As a further aspect of the present invention, based on the material classification feature values, the steps of acquiring images of the platform area using an image acquisition device, extracting texture features of the clearance marking area, identifying blurred areas of the markings by calculating the directional gradient change rate, and calculating the gradient direction interpolation of the blurred areas to obtain the clearance marking correction value are as follows:

[0022] S201: Obtain the material classification feature value, call the image acquisition device to acquire image data of the platform area, identify the boundary marker area, calculate the pixel grayscale gradient of the target area, extract local texture features according to the gradient direction distribution, and obtain boundary marker texture feature data.

[0023] S202: Based on the boundary marker texture feature data, by calculating the directional gradient change rate of adjacent pixels, identify the blurred region of the marker, mark the pixel range of the blurred region, and obtain the boundary marker blurred region data;

[0024] S203: Based on the blurred region data of the boundary marker, calculate the interpolation parameters of the pixel gradient direction within the blurred region to obtain the boundary marker correction value.

[0025] As a further aspect of the present invention, based on the clearance marker correction value, point cloud data of the platform area is acquired using lidar, and combined with image information, the platform point cloud data is spatiotemporally synchronized and registered to construct a platform structure model. The specific steps are as follows:

[0026] S301: Obtain the clearance mark correction value, call the lidar to collect point cloud data of the platform area, parse the spatial coordinate information of the point cloud data, and obtain the preliminary point cloud data of the platform;

[0027] S302: Based on the preliminary point cloud data of the platform, call the image data of the platform area, calculate the spatial coordinate matching error between the point cloud data and the image data, adjust the spatial coordinates of the point cloud data according to the alignment of the platform edge points and boundary marker points, and obtain the platform matching point cloud data;

[0028] S303: Based on the platform matching point cloud data, extract the structural morphology information of multiple areas of the platform, construct the spatial model of the platform, and obtain the platform structure model.

[0029] As a further aspect of the present invention, based on the platform structure model, the steps of calculating the spatial offset and offset angle of multiple points and generating structural offset values ​​by comparing the platform structure data with the platform structure data in the engineering design drawings are as follows:

[0030] S401: Obtain the platform structure model, call the engineering design drawings, extract the spatial coordinates of the platform edge, platform facilities, and clearance markings in the drawings, and obtain the platform reference coordinates;

[0031] S402: Based on the platform reference coordinates, by calculating the spatial position difference between the platform point cloud data and the reference coordinates, multiple spatial offset points are detected, and platform offset point data is obtained.

[0032] S403: Based on the platform offset point data, calculate the spatial offset distance of multiple offset points, analyze the offset direction angle of multiple points, and obtain the platform structure offset value.

[0033] As a further aspect of the present invention, the steps of extracting offset point information based on the structural offset value, calling the train's external dimensions and running trajectory, calculating the collision risk between the platform clearance and the train, and generating a collision risk value are as follows:

[0034] S501: Obtain the platform structure offset value, extract the spatial coordinates of multiple offset points, calibrate the spatial distribution of offset points, and obtain platform clearance offset point data.

[0035] S502: Based on the platform clearance offset point data, call the train's external dimensions and running trajectory information to calculate the space occupied by the train when passing through the platform clearance area, and obtain the train passage space data;

[0036] S503: Based on the train passage space data, the collision risk is assessed and a collision risk value is obtained by calculating the distance between the offset point and the train outline.

[0037] As a further aspect of the present invention, the specific formula for assessing collision risk is as follows:

[0038]

[0039] Calculate the collision risk coefficient;

[0040] Among them, R c d represents the collision risk factor. s Let d be the actual measured distance between the s-th offset point and the train outline. safe For the safety distance in the platform clearance design standard, v s Let t be the speed of the train as it passes through the region at the s-th offset point. s Let be the time required for the train to pass through the region of the s-th offset point, where s is the index of the s-th offset point, and S is the total number of offset points involved in the collision risk calculation.

[0041] On the other hand, a computer vision-based intelligent detection system for station platform clearance is provided. This system is applied to a computer vision-based intelligent detection method for station platform clearance, and includes:

[0042] The spectral analysis module is based on the spectral data of the platform area. By parsing the spectral reflectance value of each pixel, it calculates the spectral peak position, bandwidth half width at half maximum and spectral gradient change, calls the known material spectral database, classifies the platform material, detects obstacles, and obtains the platform material classification feature value.

[0043] The marking optimization module collects image data of the platform area based on the platform material classification feature value, extracts pixel features of the boundary marking area, identifies blurred marking areas, calculates gradient direction interpolation of the blurred areas, and obtains boundary marking correction values.

[0044] Based on the clearance marker correction value, the structure construction module calls the lidar to collect point cloud data of the platform area, extracts platform edge points, facility points and clearance marker points, calculates the matching deviation between image information and point cloud data, performs spatiotemporal synchronization registration, adjusts the spatial coordinates of point cloud data, and constructs a platform structure model.

[0045] The offset calculation module, based on the platform structure model, calls the engineering design drawing data, compares the offset of the point cloud data with the design standard, calculates the spatial offset of multiple points on the platform, and obtains the platform structure offset value.

[0046] Based on the platform structure offset value, the risk assessment module extracts the offset points within the platform clearance, calls the train's external dimensions and running trajectory, calculates the collision risk, and obtains the collision risk value.

[0047] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0048] By acquiring spectral data and comparing reflectance values ​​to identify material types, and using texture features and gradient change rate analysis to identify blurred areas, we can reduce misjudgments of blurred identifications. We adopt spatiotemporal synchronous registration of point cloud data and image information, and integrate multi-source perception information to enhance the spatiotemporal consistency of 3D reconstruction, thereby improving the dimensional accuracy of the platform model. By comparing spatial offset with engineering design drawings, we can establish a quantitative evaluation system for structural deviations, locate deformation areas, and improve the predictability of safety warnings. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0051] Figure 2 This is a system flowchart of the present invention. Detailed Implementation

[0052] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0053] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0054] In the embodiments of the present invention, the terms "image" and "picture" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same. The terms "of," "corresponding," and "corresponding" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same.

[0055] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0056] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0057] Please see Figure 1 This invention provides a technical solution: a computer vision-based intelligent detection method for station platform clearance, comprising the following steps:

[0058] S1: Acquire spectral data of the platform area, extract spectral features based on the reflectance value of each pixel within the spectral range, compare them with the features of various known materials, identify the material types at multiple locations, detect obstacles, and generate material classification feature values.

[0059] S2: Based on the material classification feature value, use image acquisition equipment to acquire images of the platform area, extract the texture features of the boundary marker area, identify the blurred area of ​​the marker by calculating the directional gradient change rate, and calculate the gradient direction interpolation of the blurred area to obtain the boundary marker correction value.

[0060] S3: Based on the clearance mark correction value, use LiDAR to acquire point cloud data of the platform area, combine with image information, perform spatiotemporal synchronization registration of the platform point cloud data, and construct a platform structure model;

[0061] S4: Based on the platform structure model, by comparing it with the platform structure data in the engineering design drawings, calculate the spatial offset and offset angle of multiple points, and generate the structural offset value.

[0062] S5: Based on the structural offset value, extract the offset point information, call the train's external dimensions and running trajectory, calculate the collision risk between the platform clearance and the train, and generate a collision risk value.

[0063] The material classification feature values ​​specifically include the material spectral reflectance parameter, material type matching result, and obstacle spectral feature value. The clearance mark correction values ​​include the gradient change rate of the clearance mark texture direction, the gradient interpolation calculation value of the blurred area, and the mark boundary correction parameter. The platform structure model specifically includes the platform edge point cloud coordinates, clearance mark point cloud data, and platform facility point cloud data. The structural offset values ​​include the change in spatial coordinates of the offset point, the change in the rotation angle of the offset point, and the platform structure offset trend data. The collision risk values ​​include the platform offset distance, the platform offset angle, and the collision risk level.

[0064] The specific steps for acquiring spectral data of the platform area, extracting spectral features based on the reflectance value of each pixel within the spectral range, comparing these features with the features of various known materials, identifying material types at multiple locations, detecting obstacles, and generating material classification feature values ​​are as follows:

[0065] S101: Acquire spectral data of the platform area, extract the reflectance value of each pixel, calculate the spectral peak position, bandwidth half width at half maximum, and spectral slope parameters of multiple bands, extract spectral features of multiple positions, and generate a set of spectral feature parameters.

[0066] Spectral data of the platform area is acquired using a spectral sensor across multiple wavelengths, covering the visible, near-infrared, and short-wave infrared regions. For each pixel, spectral reflectance values ​​are extracted, and the spectral peak positions across multiple wavelengths are calculated. The spectral peak position refers to the wavelength value corresponding to the maximum reflectance of a specific material within the spectral range. Specifically, for the spectral curves of different materials, the first derivative of the spectral intensity is calculated, and the zero intersection point is determined to obtain the spectral peak. The bandwidth at half maximum width (HWHM) parameter describes the width of the spectral peak, which is the wavelength difference at half the peak intensity of the spectral curve. The HWHM can be obtained by calculating the two wavelengths corresponding to half the peak intensity. The values ​​are obtained by calculating the differences. For example, when the peak reflectance of a material is 75%, its bandwidth (half-width at half maximum) can be determined by calculating the wavelength difference at which the reflectance is 37.5%. Additionally, the spectral slope parameter is calculated. This parameter characterizes the rate of change of the spectral curve within a specific wavelength range. Specifically, it is calculated by segmenting the spectral curve, calculating the slope of each segment, and selecting the slope values ​​for key wavelength bands. If the rate of change of the spectral curve within a certain wavelength range is large, then that wavelength band may be a key feature point for material identification. For example, for the surface of a cement platform, its rate of change of spectral reflectance in the near-infrared band is much higher than that of metallic materials, thus it can be used as a key feature parameter. The formula for calculating the spectral slope is as follows:

[0067]

[0068] Where S is the spectral slope, R1 and R2 are the reflectances at two wavelengths, and λ1 and λ2 are the corresponding wavelength positions. Assuming a material has a reflectance of 0.45 at 800 nm and 0.60 at 850 nm, its spectral slope is calculated as follows:

[0069]

[0070] Finally, spectral features from multiple locations on the platform are extracted, and a set of spectral feature parameters is generated.

[0071] S102: Based on the spectral feature parameter set, call the spectral database of known materials, calculate the spectral similarity of materials at multiple locations on the platform, calibrate the material type at multiple locations on the platform surface, and obtain the distribution data of platform material categories;

[0072] The specific formula for calculating the spectral similarity of materials at multiple locations on the platform is as follows:

[0073]

[0074] Calculate the spectral matching index;

[0075] Where S′ represents the matching degree index between the platform's measured spectrum and the known material spectrum, A k B represents the reflectance value of the spectrum measured at the k-th wavelength point on the platform. k This represents the reflectance value of the known material spectrum at the k-th wavelength point, where K represents the total number of wavelength points used to calculate the spectral matching degree, and k represents the wavelength point index in the spectral measurement.

[0076] formula:

[0077]

[0078] Detailed explanation of the formula and its calculation derivation:

[0079] The formula is used to calculate the matching degree between the measured spectrum of the platform and the known material spectrum, and the result is used to identify the material type at each location on the platform surface;

[0080] Parameter meanings and settings:

[0081] A k The reflectance value of the spectrum measured at the platform at the k-th wavelength is represented by A1 = 0.42, A2 = 0.65, and A3 = 0.81.

[0082] B k The reflectance value of the known material spectrum at the k-th wavelength is represented by B1 = 0.40, B2 = 0.70, and B3 = 0.79.

[0083] K represents the total number of wavelength points used to calculate the spectral matching degree, which is set to 3;

[0084] k represents the wavelength point index in the spectral measurement;

[0085] Substitute the parameters into the formula to calculate:

[0086]

[0087] The spectral matching degree S′ = 0.9971 indicates that the measured spectrum of the platform is highly similar to the spectrum of the known concrete material. Therefore, the material of the platform area can be identified as concrete. This value can be used as part of the platform material category distribution data to identify the material classification of the platform surface and to detect whether there are abnormal materials or obstacles in the platform area.

[0088] S103: Call the platform material category distribution data, detect spectral reflectance anomalies in the platform area, identify obstacles, record location information in real time, and generate material classification feature values;

[0089] The system calls upon the platform material category distribution data to analyze whether the spectral reflectance of each measuring point matches the material type. It detects spectral reflectance anomalies within the platform area. For pixels with reflectance values ​​significantly deviating from the standard material range, it records the changes in spectral peak value and bandwidth (half-width at half maximum). It calculates the spectral similarity between the anomaly and surrounding material points. If the spectral similarity between the anomaly and adjacent material measuring points is below a set threshold of 0.85, the point is considered a potential obstacle. Multiple consecutive anomaly points are merged using spatial clustering to form an obstacle region. Simultaneously, the system analyzes the obstacle's outline and calculates the spectral characteristic variation amplitude of the obstacle-covered area. If the spectral variation in the obstacle region is significant, such as a large abrupt change in reflectance on a glass surface or a nonlinear spectral change in a concrete area, it is further confirmed as a non-platform material. The system then marks the center position of the obstacle, records its coordinates, stores real-time detection data, and calculates the obstacle reflectance anomaly degree.

[0090]

[0091] Among them, D r R represents the reflectivity anomaly. obs R is the spectral reflectance at the current measuring point. std Let be the spectral reflectance of an adjacent known material. For example, if the spectral reflectance of a measuring point is 0.75, and the reference reflectance of the adjacent concrete material is 0.50, then the calculation is as follows:

[0092]

[0093] When D rWhen the value is greater than 0.3, the point is identified as a potential obstacle. Further obstacle contour extraction is performed to obtain material classification feature values.

[0094] Based on material classification feature values, images of the platform area are acquired using image acquisition equipment. Texture features of the clearance marking area are extracted. By calculating the directional gradient change rate, blurred areas of the markings are identified, and gradient direction interpolation of the blurred areas is calculated to obtain the clearance marking correction value. The specific steps are as follows:

[0095] S201: Obtain material classification feature values, call the image acquisition device, acquire image data of the platform area, identify the boundary marker area, calculate the pixel grayscale gradient of the target area, extract local texture features according to the gradient direction distribution, and obtain boundary marker texture feature data.

[0096] After obtaining the material classification feature values, an image acquisition device is used, and a high-resolution camera is selected to acquire images of the platform area. This ensures balanced lighting conditions and avoids overexposure or underexposure that could lead to loss of image details. Based on the acquired image data, the boundary marker area of ​​the platform is located according to the pixel intensity distribution. Color feature extraction is used to identify the boundary marker boundaries. Boundary marker colors typically have standardized characteristics; for example, yellow or red lines show a clear contrast with the surrounding background color. Further edge detection methods are used to obtain the outline information of the markers, eliminating interfering pixels in non-target areas, extracting the pixel grayscale values ​​of the target areas, and calculating the grayscale gradient to quantify the spatial changes in pixel brightness. The grayscale gradient calculation uses the Sobel operator, which obtains the gradient magnitude and direction by calculating the pixel grayscale difference in the horizontal and vertical directions. The specific calculation is as follows:

[0097]

[0098] Among them, G θ G represents the grayscale gradient magnitude of a pixel. x and G y These represent the grayscale change rates in the horizontal and vertical directions, respectively, and are calculated as follows:

[0099] G x =I(i+1,j)-I(i-1,j), G y = I(i,j+1)-I(i,j-1);

[0100] Where I(i,j) represents the grayscale value of pixel (i,j). Assuming the grayscale values ​​of a certain pixel are I(3,2) = 120, I(5,2) = 150, I(4,1) = 110, and I(4,3) = 140, then calculate:

[0101] G x =150-120=30;

[0102] G y =140-110=30;

[0103]

[0104] For pixels with a grayscale gradient greater than a set threshold of 40, they are regarded as boundary marker edge points. The gradient direction distribution is further analyzed, and the detailed texture features of the marker area are extracted using a local texture descriptor to obtain boundary marker texture feature data.

[0105] S202: Based on the boundary marker texture feature data, by calculating the directional gradient change rate of adjacent pixels, identify the blurred region of the marker, mark the pixel range of the blurred region, and obtain the boundary marker blurred region data;

[0106] Based on the boundary marker texture feature data, the directional gradient change rate of adjacent pixels is calculated to determine whether there is blurring in the boundary marker area. For adjacent pixels in the image, the rate of change of their grayscale gradient is calculated. If the rate of change is lower than a set threshold, the area may be blurry. The directional gradient change rate is calculated using the local gradient ratio method, as detailed below:

[0107]

[0108] Among them, T θ G is the rate of change of the directional gradient. θ (p) represents the grayscale gradient of pixel p, G θ (q) represents the grayscale gradient between adjacent pixels q. If the gradient change rate between adjacent pixels is less than a set threshold of 0.1, then the region is considered blurry. For example, if the grayscale gradients of two adjacent pixels in a certain region are 42.4 and 40.0 respectively, the calculation is as follows:

[0109]

[0110] Since the value is less than 0.1, the region is determined to be blurry. The connected component labeling method is then used to label all pixel regions with gradient change rates below the threshold, forming a continuous blurry region. By calculating the pixel distribution of the blurry region, the blurry boundary is delineated, and the boundary marker blurry region data is obtained.

[0111] S203: Based on the data of the blurred region of the boundary marker, calculate the interpolation parameters of the pixel gradient direction within the blurred region and obtain the boundary marker correction value;

[0112] Based on the blurred region data of boundary markers, interpolation calculations are performed on the pixel gradient directions within the blurred regions to correct the boundary information of the blurred regions and improve the recognizability of the boundary markers. The interpolation calculation adopts the bilinear interpolation method, which interpolates the gradient direction of the current blurred pixel based on the gradient direction information of adjacent pixels. The interpolation parameters are calculated as follows:

[0113]

[0114] Among them, G interp G represents the gradient direction after interpolation. θ (a), G θ (b), G θ (c), G θ (d) represents the gradient values ​​of the four neighboring directions of the pixel. Assuming the gradient values ​​of the four neighboring areas of a pixel in a blurred region are 38.0, 39.5, 40.2, and 41.0, the calculation is as follows:

[0115]

[0116] After the interpolation calculation is completed, interpolation calculation is performed on all blurred area pixels to correct their gradient direction and generate the corrected boundary marker area. By comparing the gradient distribution before and after interpolation, the boundary marker boundary is ensured to be clearly distinguishable, and the boundary marker correction value is obtained.

[0117] Based on the clearance marker correction value, point cloud data of the platform area is acquired using LiDAR, and combined with image information, the platform point cloud data is spatiotemporally synchronized and registered to construct the platform structure model. The specific steps are as follows:

[0118] S301: Obtain the clearance mark correction value, call the lidar to collect point cloud data of the platform area, parse the spatial coordinate information of the point cloud data, and obtain the preliminary point cloud data of the platform;

[0119] After obtaining the clearance marking correction value, the lidar is used to scan the platform area, collect 3D point cloud data, and analyze the spatial coordinate information in the point cloud data. The lidar emits laser pulses and records the time difference from emission to return. The spatial distance of each point is calculated based on the speed of light. The lidar measurement angle range θ is set to 180°. The spatial coordinates (x, y, z) of each measurement point in the point cloud data are calculated as follows:

[0120] x=dcos(θ)cos(φ), y=dcos(θ)sin(φ), z=dsin(θ);

[0121] Where d represents the laser ranging value, φ is the horizontal angle, and θ is the vertical angle. Assuming the laser ranging value at a certain measurement point is d = 2.5m, φ = 45°, and θ = 30°, the calculation is as follows:

[0122] x=2.5cos(30°)cos(45°)=1.77m;

[0123] y=2.5cos(30°)sin(45°)=1.77m;

[0124] z = 2.5sin(30°) = 1.25m;

[0125] The spatial coordinates of all points are calculated sequentially to establish preliminary platform point cloud data. Noise is removed from the point cloud data, and the distance between adjacent points is calculated. If the distance between adjacent points is greater than a set threshold of 0.5m, the point is removed to eliminate abnormal points and obtain preliminary platform point cloud data.

[0126] S302: Based on the preliminary point cloud data of the platform, call the image data of the platform area, calculate the spatial coordinate matching error between the point cloud data and the image data, adjust the spatial coordinates of the point cloud data according to the alignment of the platform edge points and boundary marker points, and obtain the platform matching point cloud data;

[0127] Based on the initial point cloud data of the platform, image data of the platform area is retrieved, and coordinate projection is performed on each point in the point cloud data. The point cloud data is projected onto the image plane and compared with the platform edge points and boundary marker points in the image. The positional deviation between the projected point cloud points and the actual image points is calculated. The deviation is calculated using pixel distance, corresponding to the actual error value in the physical world. For points with large errors, their relative error distribution is analyzed to determine whether the offset is caused by the laser scanning angle during point cloud data acquisition, surface material reflection characteristics, or environmental occlusion. If a systematic offset exists, coordinate adjustments are made according to the overall trend of the point cloud, and local interpolation corrections are performed on the error points. The spatial coordinates of the platform edge points are calculated, and the boundary position of the point cloud data is adjusted using the platform boundary information in the image to ensure that the boundary lines in the point cloud data are aligned with the boundary lines extracted from the image data. The adjustment of the boundary marker points is based on the boundary markers in the image data. The actual offset of the marker points in the point cloud data is calculated, and a weighted smoothing method is used for global coordinate correction to reduce the overall error of the point cloud, enabling the point cloud data to more accurately reflect the true spatial layout of the platform. Ultimately, the adjusted point cloud data will be used to construct the subsequent platform structure model, ensuring the spatial accuracy of the platform point cloud data, improving the accuracy of platform clearance measurement, and obtaining matching point cloud data for the platform.

[0128] S303: Based on the platform matching point cloud data, extract the structural morphology information of multiple areas of the platform, construct the spatial model of the platform, and obtain the platform structural model;

[0129] Based on the platform matching point cloud data, structural morphology information of multiple areas of the platform is extracted, the platform structural features are divided into regions, and the point cloud density of different regions is calculated. The point cloud density D is calculated as follows:

[0130]

[0131] Where, N p Let A be the number of point clouds per unit area, and A be the area of ​​the region. Assume the area of ​​a certain platform region is A = 4m². 2 The number of point clouds N in a unit area p =12000, then the calculation is as follows:

[0132]

[0133] If the point cloud density is lower than the set threshold of 2000 points / m 2 If necessary, the point cloud data of the region is supplemented to improve the integrity of the model. Point cloud fitting is performed on each region to establish structural surface models of the platform ground, edges, columns, etc., and finally the platform structure model is obtained.

[0134] Based on the platform structure model, the steps for calculating the spatial offset and offset angle of multiple points and generating the structural offset value are as follows: By comparing the platform structure data with the engineering design drawings, the specific steps are:

[0135] S401: Obtain the platform structure model, call the engineering design drawings, extract the spatial coordinates of the platform edge, platform facilities, and clearance markings in the drawings, and obtain the platform reference coordinates;

[0136] After obtaining the platform structure model, the engineering design drawings are retrieved, and the spatial coordinate information of the platform edges, platform facilities, and clearance markings in the drawings is analyzed. Multiple key point coordinates are extracted from the design drawings, and using a three-dimensional coordinate representation, with the platform origin O(0,0,0) as the reference, the platform edge point P is set. e (x e ,y e ,z e Platform facility point P s (x s ,y s ,z s ), boundary marker point P l (x l ,y l ,z l To generate baseline coordinate data, calculate the Euclidean distance between each point using the following formula:

[0137]

[0138] Where, d ab The spatial distance between two points, (x a ,y a ,z a ) and (x b ,yb ,z b Let P be the coordinates of two points, and let P be the edge point of the platform. e (10,5,2) and boundary marker P l (12,8,2), then the calculation is as follows:

[0139]

[0140] Calculate the coordinates of all key points sequentially, establish the platform reference coordinate data, and store it for reference in subsequent comparative analysis to obtain the platform reference coordinates.

[0141] S402: Based on the platform reference coordinates, multiple spatial offset points are detected by calculating the spatial position difference between the platform point cloud data and the reference coordinates, and the platform offset point data is obtained.

[0142] Based on the platform's reference coordinates, the platform's point cloud data is retrieved, and the spatial position difference between the point cloud data and the reference coordinates is calculated. Identical platform edge points, platform facility points, and clearance marker points are selected from the point cloud data, and their corresponding coordinates P′ are calculated. e (x′ e ,y′ e ,z′ e ), P′ s (x′ s ,y′ s ,z′ s ), P′ l (x′ l ,y′ l ,z′ l ), calculate the offset Δd of each point from the reference coordinate:

[0143]

[0144] Assume the detection coordinates of a certain platform edge point are P′ e (10.5, 5.8, 2.2), the corresponding reference coordinates are P. e (10,5,2), the offset is calculated as follows:

[0145]

[0146] If the offset is greater than the set threshold of 0.5m, then record the point as the offset point, filter multiple platform offset points, and obtain the platform offset point data.

[0147] S403: Based on the platform offset point data, calculate the spatial offset distance of multiple offset points, analyze the offset direction angle of multiple points, and obtain the platform structure offset value.

[0148] Based on the platform offset point data, the spatial offset distance of multiple offset points is calculated, and the offset direction angle of multiple points is analyzed. First, the offset direction angle θ of each offset point is calculated. The formula for calculating the direction angle is as follows:

[0149]

[0150] Assume the detection coordinates of a certain station offset point are P′ e (11,6,2), the corresponding reference coordinate is P. e (10,5,2), calculate the direction angle:

[0151]

[0152] Calculate the orientation angles of all offset points and classify them into longitudinal offset, lateral offset, and height offset. Based on the orientation angle range of 0°-45°, they are classified as lateral offset and 45°-90° as longitudinal offset. Record the offset direction data of all offset points to obtain the platform structure offset value.

[0153] Based on the structural offset value, the offset point information is extracted. The train's external dimensions and running trajectory are retrieved to calculate the collision risk between the platform clearance and the train. The specific steps for generating the collision risk value are as follows:

[0154] S501: Obtain the platform structure offset value, extract the spatial coordinates of multiple offset points, calibrate the spatial distribution of offset points, and obtain the platform clearance offset point data.

[0155] After obtaining the platform structure offset values, the spatial coordinate information of multiple offset points is extracted. The positions of the offset points are represented using a three-dimensional coordinate system, and each offset point is set as P. i (x i ,y i ,z i The system calculates the spatial distribution of these points within the platform clearance and defines the centroid G(x) of the clearance offset points by calculating the geometric centers of all offset points. g ,y g ,z g ):

[0156]

[0157] Where N is the total number of offset points, (x i ,y i ,z i Let P(10,5,2), P2(12,6,2.5), and P3(11,5.5,2.2) be the coordinates of the i-th offset point. Assuming the detected offset points are P1(10,5,2), P2(12,6,2.5), and P3(11,5.5,2.2), then the following can be calculated:

[0158]

[0159] Calculate the spatial distribution of all offset points and record the coordinates of the clearance offset points to obtain the platform clearance offset point data.

[0160] S502: Based on the platform clearance offset point data, call the train's external dimensions and running trajectory information to calculate the space occupied by the train when passing through the platform clearance area and obtain the train's passage space data;

[0161] Based on the platform clearance offset point data, and by calling the train's external dimensions and trajectory information, a train passage space model is established, with the train's external dimensions set as W. t H t ,L t (representing the train's width, height, and length respectively), the train's trajectory is determined by the track centerline C(x). c ,y c Defined as follows: calculates the area occupied by a train when passing through the platform clearance zone, and sets the maximum outer coordinate P when the train reaches the center position of the platform. t (x t ,y t ,z t ), calculate the train envelope range within the platform clearance area, taking the train width as an example, calculate the maximum horizontal extension from the track center point to the train side boundary:

[0162]

[0163] Assuming the train width W t =3m, track center position x c =10m, then:

[0164]

[0165] The maximum outer perimeter of the train in the platform area is calculated sequentially to obtain train passage space data.

[0166] S503: Based on train passage space data, the collision risk is assessed and a collision risk value is obtained by calculating the distance between the offset point and the train outline.

[0167] The specific formula for assessing collision risk is:

[0168]

[0169] Calculate the collision risk coefficient;

[0170] Among them, R c d represents the collision risk factor. s Let d be the actual measured distance between the s-th offset point and the train outline. safe For the safety distance in the platform clearance design standard, v sLet t be the speed of the train as it passes through the region at the s-th offset point. s Let be the time required for the train to pass through the region of the s-th offset point, where s is the index of the s-th offset point, and S is the total number of offset points involved in the collision risk calculation.

[0171] formula:

[0172]

[0173] Detailed explanation of the formula and its calculation derivation:

[0174] The formula is used to calculate the collision risk coefficient between the platform clearance offset point and the train, and the result is used to assess the safety impact of platform structural deformation or facility offset on train operation.

[0175] Parameter meanings and settings:

[0176] R c The collision risk factor reflects the safety of the platform offset point and the train operating space.

[0177] d s The actual measured distances between the s-th platform offset point and the train outline are set to 0.48m, 0.55m, 0.62m, 0.51m, and 0.45m, respectively.

[0178] d safe The safety reference value for the platform clearance is set at 0.5m;

[0179] v s The speed of the train passing through this offset point area is set to 15 m / s;

[0180] t s Let t be the time required for the train to pass through the offset point area. s =0.5s;

[0181] S represents the total number of platform offset points involved in the calculation, which is set to 5;

[0182] Substitute the parameters into the formula to calculate:

[0183]

[0184]

[0185] The results indicate that the collision risk coefficient of the current platform clearance offset point is 2.25, which means that the offset point has a high collision risk and structural adjustments or clearance correction measures are required.

[0186] Please see Figure 2A computer vision-based intelligent detection system for station platform clearance, used to execute the aforementioned computer vision-based intelligent detection method for station platform clearance, the system comprising:

[0187] The spectral analysis module is based on the spectral data of the platform area. By parsing the spectral reflectance value of each pixel, it calculates the spectral peak position, bandwidth half width at half maximum and spectral gradient change, calls the known material spectral database, classifies the platform material, detects obstacles, and obtains the platform material classification feature value.

[0188] The marking optimization module collects image data of the platform area based on the platform material classification feature value, extracts pixel features of the boundary marking area, identifies blurred marking areas, calculates gradient direction interpolation of the blurred areas, and obtains boundary marking correction values.

[0189] The structure building module, based on the clearance marker correction value, calls the lidar to collect point cloud data of the platform area, extracts platform edge points, facility points and clearance marker points, calculates the matching deviation between image information and point cloud data, performs spatiotemporal synchronous registration, adjusts the spatial coordinates of point cloud data, and builds a platform structure model.

[0190] The offset calculation module is based on the platform structure model, calls the engineering design drawing data, compares the offset of point cloud data with the design standard, calculates the spatial offset of multiple points on the platform, and obtains the platform structure offset value.

[0191] The risk assessment module extracts the offset points within the platform clearance based on the platform structure offset value, calls the train's external dimensions and running trajectory, calculates the collision risk, and obtains the collision risk value.

[0192] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0193] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0194] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0195] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0196] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0197] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0198] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0199] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0200] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0201] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0202] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A computer vision-based intelligent detection method for station platform clearance, characterized in that, The method includes: S1: Acquire spectral data of the platform area, extract spectral features based on the reflectance value of each pixel within the spectral range, compare them with the features of various known materials, identify the material types at multiple locations, detect obstacles, and generate material classification feature values. S2: Based on the material classification feature value, use an image acquisition device to acquire an image of the platform area, extract the texture features of the boundary marker area, identify the blurred area of ​​the marker by calculating the directional gradient change rate, and calculate the gradient direction interpolation of the blurred area to obtain the boundary marker correction value. S3: Based on the clearance mark correction value, use lidar to acquire point cloud data of the platform area, combine with image information, perform spatiotemporal synchronization registration of the platform point cloud data, and construct a platform structure model. S4: Based on the platform structure model, by comparing it with the platform structure data in the engineering design drawings, calculate the spatial offset and offset angle of multiple points, and generate the structural offset value; S5: Based on the structural offset value, extract the offset point information, call the train's outer dimensions and running trajectory, calculate the collision risk between the platform clearance and the train, and generate a collision risk value.

2. The intelligent detection method for station platform clearance based on computer vision according to claim 1, characterized in that, The material classification feature values ​​specifically include material spectral reflectance parameters, material type matching results, and obstacle spectral feature values. The clearance marker correction values ​​include the clearance marker texture direction gradient change rate, the gradient interpolation calculation value of the blurred area, and the marker boundary correction parameters. The platform structure model specifically includes the platform edge point cloud coordinates, clearance marker point cloud data, and platform facility point cloud data. The structure offset values ​​include the change in spatial coordinates of the offset point, the change in the rotation angle of the offset point, and the platform structure offset trend data. The collision risk values ​​include the platform offset distance, the platform offset angle, and the collision risk level.

3. The intelligent detection method for station platform clearance based on computer vision according to claim 1, characterized in that, The specific steps for acquiring spectral data of the platform area, extracting spectral features based on the reflectance value of each pixel within the spectral range, comparing these features with the features of various known materials, identifying material types at multiple locations, detecting obstacles, and generating material classification feature values ​​are as follows: S101: Acquire spectral data of the platform area, extract the reflectance value of each pixel, calculate the spectral peak position, bandwidth half width at half maximum, and spectral slope parameters of multiple bands, extract spectral features of multiple positions, and generate a set of spectral feature parameters. S102: Based on the spectral feature parameter set, call the spectral database of known materials, calculate the spectral similarity of materials at multiple locations on the platform, calibrate the material type at multiple locations on the platform surface, and obtain the platform material category distribution data; S103: Call the platform material category distribution data, detect spectral reflectance anomalies in the platform area, identify obstacles, record location information in real time, and generate material classification feature values.

4. The intelligent detection method for station platform clearance based on computer vision according to claim 3, characterized in that, The specific formula for calculating the spectral similarity of materials at multiple locations on the platform is as follows: Calculate the spectral matching index; Where S′ represents the matching degree index between the platform's measured spectrum and the known material spectrum, A k B represents the reflectance value of the spectrum measured at the k-th wavelength point on the platform. k This represents the reflectance value of the known material spectrum at the k-th wavelength point, where K represents the total number of wavelength points used to calculate the spectral matching degree, and k represents the wavelength point index in the spectral measurement.

5. The intelligent detection method for station platform clearance based on computer vision according to claim 1, characterized in that, Based on the material classification feature values, the following steps are taken to obtain the platform area image using an image acquisition device, extract the texture features of the clearance marking area, identify blurred areas of the markings by calculating the directional gradient change rate, and calculate the gradient direction interpolation of the blurred areas to obtain the clearance marking correction value: S201: Obtain the material classification feature value, call the image acquisition device to acquire image data of the platform area, identify the boundary marker area, calculate the pixel grayscale gradient of the target area, extract local texture features according to the gradient direction distribution, and obtain boundary marker texture feature data. S202: Based on the boundary marker texture feature data, by calculating the directional gradient change rate of adjacent pixels, identify the blurred region of the marker, mark the pixel range of the blurred region, and obtain the boundary marker blurred region data; S203: Based on the blurred region data of the boundary marker, calculate the interpolation parameters of the pixel gradient direction within the blurred region to obtain the boundary marker correction value.

6. The intelligent detection method for station platform clearance based on computer vision according to claim 1, characterized in that, Based on the clearance marker correction value, the following steps are taken to acquire point cloud data of the platform area using lidar, and combine it with image information to perform spatiotemporal synchronization registration of the platform point cloud data to construct the platform structure model: S301: Obtain the clearance mark correction value, call the lidar to collect point cloud data of the platform area, parse the spatial coordinate information of the point cloud data, and obtain the preliminary point cloud data of the platform; S302: Based on the preliminary point cloud data of the platform, call the image data of the platform area, calculate the spatial coordinate matching error between the point cloud data and the image data, adjust the spatial coordinates of the point cloud data according to the alignment of the platform edge points and boundary marker points, and obtain the platform matching point cloud data; S303: Based on the platform matching point cloud data, extract the structural morphology information of multiple areas of the platform, construct the spatial model of the platform, and obtain the platform structure model.

7. The intelligent detection method for station platform clearance based on computer vision according to claim 1, characterized in that, Based on the platform structure model, the steps for calculating the spatial offset and offset angle of multiple points and generating the structural offset value by comparing it with the platform structure data in the engineering design drawings are as follows: S401: Obtain the platform structure model, call the engineering design drawings, extract the spatial coordinates of the platform edge, platform facilities, and clearance markings in the drawings, and obtain the platform reference coordinates; S402: Based on the platform reference coordinates, by calculating the spatial position difference between the platform point cloud data and the reference coordinates, multiple spatial offset points are detected, and platform offset point data is obtained. S403: Based on the platform offset point data, calculate the spatial offset distance of multiple offset points, analyze the offset direction angle of multiple points, and obtain the platform structure offset value.

8. The intelligent detection method for station platform clearance based on computer vision according to claim 1, characterized in that, Based on the structural offset value, the offset point information is extracted, the train's external dimensions and running trajectory are retrieved, and the collision risk between the platform clearance and the train is calculated. The specific steps for generating the collision risk value are as follows: S501: Obtain the platform structure offset value, extract the spatial coordinates of multiple offset points, calibrate the spatial distribution of offset points, and obtain platform clearance offset point data. S502: Based on the platform clearance offset point data, call the train's external dimensions and running trajectory information to calculate the space occupied by the train when passing through the platform clearance area, and obtain the train passage space data; S503: Based on the train passage space data, the collision risk is assessed and a collision risk value is obtained by calculating the distance between the offset point and the train outline.

9. The intelligent detection method for station platform clearance based on computer vision according to claim 8, characterized in that, The specific formula for assessing collision risk is as follows: Calculate the collision risk coefficient; Among them, R c d represents the collision risk factor. s Let d be the actual measured distance between the s-th offset point and the train outline. safe For the safety distance in the platform clearance design standard, v s Let t be the speed of the train as it passes through the region at the s-th offset point. s Let be the time required for the train to pass through the region of the s-th offset point, where s is the index of the s-th offset point, and S is the total number of offset points involved in the collision risk calculation.

10. A computer vision-based intelligent detection system for station platform clearance, characterized in that, The intelligent detection method for station platform clearance based on computer vision according to any one of claims 1-9, the system comprising: The spectral analysis module is based on the spectral data of the platform area. By parsing the spectral reflectance value of each pixel, it calculates the spectral peak position, bandwidth half width at half maximum and spectral gradient change, calls the known material spectral database, classifies the platform material, detects obstacles, and obtains the platform material classification feature value. The marking optimization module collects image data of the platform area based on the platform material classification feature value, extracts pixel features of the boundary marking area, identifies blurred marking areas, calculates gradient direction interpolation of the blurred areas, and obtains boundary marking correction values. Based on the clearance marker correction value, the structure construction module calls the lidar to collect point cloud data of the platform area, extracts platform edge points, facility points and clearance marker points, calculates the matching deviation between image information and point cloud data, performs spatiotemporal synchronization registration, adjusts the spatial coordinates of point cloud data, and constructs a platform structure model. The offset calculation module, based on the platform structure model, calls the engineering design drawing data, compares the offset of the point cloud data with the design standard, calculates the spatial offset of multiple points on the platform, and obtains the platform structure offset value. Based on the platform structure offset value, the risk assessment module extracts the offset points within the platform clearance, calls the train's external dimensions and running trajectory, calculates the collision risk, and obtains the collision risk value.

Images