Wheel rail state monitoring method based on machine vision

Abstract

The invention discloses a wheel rail state monitoring method based on machine vision, and relates to the technical field of rail traffic, the wheel rail state monitoring method comprises the following steps: using a light source camera to obtain wheel rail images of a wheel and a steel rail area, wheel and steel rail contour information in the wheel rail image is extracted based on a frame difference method, and the wheel center point and the steel rail top position are judged; and analyzing the distance between the central point of the wheel and the rail top position of the steel rail, analyzing the actual lifting amount of the wheel by using the distance result and the coordinate conversion relationship, monitoring the state of the wheel rail according to the actual lifting amount of the wheel, and judging the risk of derailing of the wheel. According to the vehicle wheel-rail state monitoring method provided by the invention, based on a monocular machine vision image technology, the picture of the wheel and wheel-rail contact area is analyzed to obtain the contour information of the wheel and the steel rail, and the change of the height from the center of the wheel to the rail surface is detected based on the contour information; therefore, early warning and alarming can be carried out after the train has behaviors such as rail climbing and derailing.

Description

Technical Field

[0001] The present invention relates to the field of rail transportation technology, and in particular to a wheel-rail state monitoring method based on machine vision. Background Art

[0002] Rail transit refers to a type of transportation tool or transportation system that runs on specific tracks. As rail transit develops, the safety of train operation has received more and more attention. Vehicle derailment, as a major safety hazard of rail transit, not only causes huge economic losses, but also poses a serious threat to the safety of passengers' lives. Therefore, accurate monitoring of the wheel-rail status of vehicles and accurate prediction and alarm of derailment accidents are of great significance.

[0003] In recent years, with the development of rail transit safety monitoring technology, technologies and products related to derailment monitoring have been put into use. The traditional train derailment monitoring device on the market is based on the principle of wheel-rail (wheel and sleeper, fasteners, etc.) impact vibration, and judges the derailment status by detecting impact and time. This method has problems such as complex judgment logic, low accuracy, and poor timeliness. In contrast, detecting the wheel lift and the relative lateral displacement of the wheel and rail is a more intuitive and accurate method to judge the derailment status, and can also achieve early warning.

[0004] For example, patent number CN117649437B discloses a method and device for measuring the lateral displacement of a rail vehicle wheel and rail. The method includes obtaining an original vehicle wheel-rail contact image set and marking multiple key point information; inputting the image set with multiple key point information labels into a key point prediction network to obtain the predicted position of each key point of vehicle wheel-rail contact, the predicted displacement deviation, and the predicted distance between two specified key points, which can improve the speed and accuracy of wheel-rail lateral displacement measurement.

[0005] At the same time, patent number CN114368411B discloses a monitoring and early warning method for train derailment safety, which uses a monitoring and early warning method based on visual behavior to analyze pictures of the wheel-rail contact area. It is more direct, the laser line positioning is accurate, the wheel-rail contact contour in the captured picture is clear, the target extraction steps are reduced, the tracking and positioning speed is fast, the calculation accuracy is high, and the early warning efficiency and accuracy are improved.

[0006] However, both of the above patents realize the lateral relative displacement of the wheel and rail, but cannot detect the amount of wheel lifting, making it impossible to issue early warnings and alarms after the train climbs the track or derails, reducing the safety of the train when in use.

[0007] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0008] In view of the problems in the related art, the present invention proposes a wheel-rail state monitoring method based on machine vision to overcome the above technical problems existing in the existing related art.

[0009] To this end, the specific technical solution adopted by the present invention is as follows:

[0010] A wheel-rail condition monitoring method based on machine vision, the wheel-rail condition monitoring method comprising:

[0011] The wheel-rail image of the wheel and rail area is obtained by using a light source camera, and the wheel and rail contour information in the wheel-rail image is extracted based on the frame difference method to determine the position of the wheel center point and the rail top;

[0012] Analyze the distance between the center point of the wheel and the top of the rail, use the distance result and coordinate transformation relationship to analyze the actual wheel lift, and monitor the state of the wheel and rail based on the actual wheel lift to determine the risk of wheel derailment.

[0013] Preferably, obtaining wheel-rail images of the wheel and rail area by using a light source camera, and extracting wheel and rail profile information in the wheel-rail image based on a frame difference method to determine the position of the wheel center point and the rail top includes:

[0014] A laser light source camera installed at the end of the axle is used to project light to obtain a wheel-rail image covering the wheel and rail area;

[0015] The frame difference method is used to perform filtering enhancement and channel feature extraction on the wheel-rail image, and a contour image is generated based on the extraction result. The contour image is then used to extract the wheel and rail contour information in the wheel-rail image.

[0016] The equation information of the wheel and rail in the image coordinate system is analyzed based on the wheel and rail profile information, and the position of the wheel center point and the rail top is approximately analyzed based on the equation information results.

[0017] Preferably, applying the frame difference method to perform filtering enhancement processing and channel feature extraction on the wheel-rail image, generating a contour image based on the extraction result, and extracting wheel and rail contour information in the wheel-rail image using the contour image includes:

[0018] Applying the frame difference method to perform gradient enhancement processing on the wheel-rail image to generate a guide image, performing filter enhancement processing on the wheel-rail image based on the guide image and the filter, and outputting the initial image according to the processing result;

[0019] Extract channel features from the primary image to obtain a secondary image, perform filtering and enhancement processing on the secondary image, and construct a pixel mean image based on the enhanced secondary image;

[0020] A pixel value subtraction operation is performed on the secondary image and the pixel mean image to obtain a contour image, and the wheel contour and rail contour are analyzed based on the pixel difference between the contour image and the initial image.

[0021] Preferably, analyzing equation information of the wheel and the rail in the image coordinate system based on the wheel and rail profile information, and approximately analyzing the position of the wheel center point and the rail top according to the equation information result includes:

[0022] Generate a vehicle image coordinate system based on the wheel profile information, define three groups of wheel coordinate points outside the outer circumference of the wheel based on the vehicle image coordinate system, and obtain the coordinates of the wheel center point based on the wheel coordinate points and the circle equation information;

[0023] Generate a rail image coordinate system based on the rail profile information, and define two groups of rail coordinate points on the rail top based on the rail image coordinate system;

[0024] The straight line equation of the rail top in the image coordinate system is obtained according to the rail coordinate points and the straight line equation, and the straight line equation of the rail top is used to approximately represent the position of the rail top in the wheel-rail image.

[0025] Preferably, analyzing the distance between the center point of the wheel and the top position of the rail, analyzing the actual wheel lift amount using the distance result and the coordinate conversion relationship, and monitoring the state of the wheel and rail according to the actual wheel lift amount, and judging the risk of wheel derailment includes:

[0026] In the image coordinate system, a distance calculation formula is generated according to the straight line equation between the wheel center point and the rail top, and the distance between the wheel center point and the rail top position is calculated using the distance calculation formula;

[0027] Determine the light source camera calibration process based on the camera's internal and external parameters, and use the calibration process results to find the coordinate projection relationship to obtain the coordinate conversion relationship between the image coordinate system and the world coordinate system;

[0028] The distance result is converted into the distance in the world coordinate based on the coordinate conversion relationship to obtain the actual distance between the center point of the wheel and the top position of the rail, and the difference between the actual distance and the preset distance is calculated;

[0029] The difference result is used as the actual wheel lift, and the state of the wheel-rail is judged based on the actual wheel lift. The risk of wheel derailment is monitored based on the wheel-rail state.

[0030] Preferably, the distance calculation formula is expressed as:

[0031]

[0032] Wherein, d represents the distance between the wheel center point and the rail top position, A represents the direction rate of the rail top straight line equation, B represents the inclination rate of the rail top straight line equation, C represents the constant term, x0 represents the horizontal coordinate of the wheel center point, and y0 represents the vertical coordinate of the wheel center point.

[0033] Preferably, determining a light source camera calibration process based on internal and external parameters of the camera, and using the calibration process result to find a coordinate projection relationship to obtain a coordinate conversion relationship between an image coordinate system and a world coordinate system includes:

[0034] The pixel size, focal length and optical axis center of the light source camera are used as the camera internal parameters, and the rotation matrix and translation matrix are generated based on the geometric relationship of the light source camera imaging to calculate the camera external parameters;

[0035] Convert the coordinate point from the world coordinate system to the camera coordinate system, obtain the coordinate of the coordinate point in the camera coordinate system, and convert the three-dimensional camera coordinate to the two-dimensional image plane;

[0036] The projection formula is defined using the camera internal parameters to project the coordinate points in the two-dimensional image plane, and the projection relationship from the three-dimensional world coordinates to the two-dimensional image coordinates is analyzed based on the projection results;

[0037] The camera calibration process is output according to the projection relationship, and the coordinate transformation relationship between the image coordinate system and the world coordinate system is obtained based on the camera calibration process.

[0038] Preferably, the projection formula is expressed as:

[0039]

[0040] In the formula, λ represents the coordinate transformation coefficient, represents the coordinate point of the image coordinate system, E represents the internal parameters of the camera, R represents the rotation matrix, t represents the translation matrix, Represents the coordinate point in the world coordinate system.

[0041] Preferably, the difference result is used as the actual wheel lift, and the wheel-rail state is judged according to the actual wheel lift. The risk of wheel derailment based on the wheel-rail state monitoring includes:

[0042] The difference between the actual distance and the preset distance is used as the actual wheel lift, and the contact state between the wheel and the rail is determined based on the actual wheel lift;

[0043] defining a position of a rolling circle and a radius of the rolling circle according to the contact state, and defining a wheel rim height based on a distance between the position of the rolling circle and a top position of the wheel;

[0044] Obtain the difference between the actual distance and the rolling circle radius, compare the difference result with the wheel flange height to determine the state of the wheel and rail, and monitor the risk of wheel derailment based on the wheel and rail state.

[0045] Preferably, when the difference between the actual distance and the rolling circle radius is greater than or equal to the wheel rim height, it indicates that the wheel has climbed the rail and there is a risk of derailment.

[0046] The beneficial effects of the present invention are:

[0047] 1. A vehicle wheel-rail state monitoring method proposed in the present invention analyzes images of wheels and wheel-rail contact areas based on monocular machine vision image technology, obtains contour information of wheels and rails based on the analysis results, and detects changes in the height of the wheel center from the rail surface based on the contour information, thereby calculating the lift of the wheel relative to the rail, so that early warning and alarm can be issued after the train climbs the rails, derails, etc., thereby improving the safety of the train when in use.

[0048] 2. The present invention uses the frame difference method to perform gradient enhancement and filter enhancement processing on the wheel-rail image, and analyzes the wheel profile and rail profile based on the pixel difference between the processed profile image and the initial image, which facilitates the subsequent measurement of the rail surface lift and lays the foundation for subsequent rail-wheel status monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 is a flow chart of a wheel-rail state monitoring method based on machine vision according to an embodiment of the present invention;

[0051] Figure 2 This is one of the schematic diagrams of a wheel lift amount detection method in a wheel-rail condition monitoring method based on machine vision according to an embodiment of the present invention;

[0052] Figure 3 This is a second schematic diagram of a method for detecting wheel lift amount in a wheel-rail condition monitoring method based on machine vision according to an embodiment of the present invention;

[0053] Figure 4 is one of the schematic diagrams of wheel-rail contact state in a wheel-rail state monitoring method based on machine vision according to an embodiment of the present invention;

[0054] Figure 5This is the second schematic diagram of the wheel-rail contact state in a wheel-rail state monitoring method based on machine vision according to an embodiment of the present invention. DETAILED DESCRIPTION

[0055] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments. They can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and advantages of the present invention.

[0056] According to an embodiment of the present invention, a wheel-rail condition monitoring method based on machine vision is provided.

[0057] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to the wheel-rail state monitoring method based on machine vision according to an embodiment of the present invention, the wheel-rail state monitoring method includes:

[0058] Step S1, using a light source camera to obtain a wheel-rail image of the wheel and rail area, and extracting wheel and rail contour information in the wheel-rail image based on a frame difference method to determine the position of the wheel center point and the rail top;

[0059] Step S2, analyzing the distance between the center point of the wheel and the top position of the rail, analyzing the actual wheel lift amount using the distance result and the coordinate conversion relationship, and monitoring the wheel-rail state based on the actual wheel lift amount to determine the risk of wheel derailment.

[0060] In this embodiment, when a wheel-rail image of the wheel and rail area is obtained by using a light source camera, and the wheel and rail contour information in the wheel-rail image is extracted based on the frame difference method to determine the wheel center point and the rail top position, a laser light source camera installed at the end position of the axle can be used to project light to obtain a wheel-rail image covering the wheel and rail area; the frame difference method is used to perform filtering enhancement processing and channel feature extraction on the wheel-rail image, a contour image is generated based on the extraction result, and the wheel and rail contour information in the wheel-rail image is extracted using the contour image; the equation information of the wheel and rail in the image coordinate system is analyzed based on the wheel and rail contour information, and the wheel center point and the rail top position are approximately analyzed based on the equation information results.

[0061] In this embodiment, when the frame difference method is applied to perform filtering enhancement processing and channel feature extraction on the wheel-rail image, a contour image is generated based on the extraction result, and the contour image is used to extract the wheel and rail contour information in the wheel-rail image, the frame difference method can be applied to perform gradient enhancement processing on the wheel-rail image to generate a guide image, the wheel-rail image is filtered and enhanced based on the guide image and the filter, and the primary image is output according to the processing result; channel feature extraction is performed on the primary image to obtain a secondary image, and filtering enhancement processing is performed on the secondary image, and a pixel mean image is constructed based on the enhanced secondary image; pixel value subtraction operation is performed on the secondary image and the pixel mean image to obtain a contour image, and the wheel contour and rail contour are analyzed based on the pixel difference between the contour image and the initial image.

[0062] In this embodiment, when analyzing the equation information of the wheel and rail in the image coordinate system based on the wheel and rail contour information, and approximating the position of the wheel center point and the rail top according to the equation information results, a vehicle image coordinate system can be generated based on the wheel contour information, and three groups of wheel coordinate points outside the outer circumference of the wheel can be defined based on the vehicle image coordinate system, and the coordinates of the wheel center point are obtained according to the wheel coordinate points and the circle equation information; a rail image coordinate system is generated based on the rail contour information, and two groups of rail coordinate points at the rail top are defined based on the rail image coordinate system; the straight line equation of the rail top in the image coordinate system is obtained according to the rail coordinate points and the straight line equation, and the position of the rail top in the wheel-rail image is approximately represented by the straight line equation of the rail top.

[0063] In this embodiment, when analyzing the distance between the wheel center point and the top of the rail, analyzing the actual wheel lift using the distance result and the coordinate conversion relationship, monitoring the state of the wheel-rail based on the actual wheel lift, and judging the risk of wheel derailment, a distance calculation formula can be generated in the image coordinate system based on the straight line equation between the wheel center point and the top of the rail, and the distance calculation formula is used to calculate the distance between the wheel center point and the top of the rail; determining the light source camera calibration process based on internal and external parameters of the camera, and using the calibration process result to find the coordinate projection relationship to obtain the coordinate conversion relationship between the image coordinate system and the world coordinate system; converting the distance result into the distance in the world coordinate based on the coordinate conversion relationship to obtain the actual distance between the wheel center point and the top of the rail, and performing a difference calculation between the actual distance and the preset distance; using the difference result as the actual wheel lift, judging the state of the wheel-rail based on the actual wheel lift, and monitoring the risk of wheel derailment based on the wheel-rail state.

[0064] In this embodiment, when determining the light source camera calibration process based on the internal and external parameters of the camera and using the results of the calibration process to find the coordinate projection relationship to obtain the coordinate transformation relationship between the image coordinate system and the world coordinate system, the pixel size, focal length and optical axis center of the light source camera can be used as the camera internal parameters, and the rotation matrix and translation matrix are generated based on the geometric relationship of the light source camera imaging to calculate the camera external parameters; the coordinate points are transformed from the world coordinate system to the camera coordinate system to obtain the coordinates of the coordinate points in the camera coordinate system, and the three-dimensional camera coordinates are transformed to the two-dimensional image plane; the projection formula is defined using the camera internal parameters in the two-dimensional image plane to project the coordinate points, and the projection relationship from the three-dimensional world coordinates to the two-dimensional image coordinates is analyzed based on the projection results; the camera calibration process is output according to the projection relationship, and the coordinate transformation relationship between the image coordinate system and the world coordinate system is obtained based on the camera calibration process.

[0065] In this embodiment, when the difference result is used as the actual wheel lift, and the state of the wheel-rail is judged based on the actual wheel lift, and the risk of wheel derailment is monitored based on the wheel-rail state, the difference result between the actual distance and the preset distance can be used as the actual wheel lift, and the contact state between the wheel and the rail is judged based on the actual wheel lift; the position of the rolling circle and the radius of the rolling circle are defined according to the contact state, and the rim height is defined based on the distance between the rolling circle position and the top position of the wheel; the difference between the actual distance and the rolling circle radius is obtained, the difference result is compared with the rim height to judge the state of the wheel-rail, and the risk of wheel derailment is monitored based on the wheel-rail state.

[0066] This embodiment discloses a vehicle wheel-rail state monitoring method, which is based on monocular machine vision image technology, analyzes images of the wheel and the wheel-rail contact area, detects the change in the height of the wheel center from the rail surface, and thus calculates the lift of the wheel relative to the rail. The specific implementation steps are as follows:

[0067] Step 1: Install a camera containing a laser light source at the end of the axle so that the camera and the wheel are relatively stationary in the x, y, and z directions. The laser light source projects light from one side of the wheel to cover the wheel and part of the rail area, and the camera captures the area to obtain a wheel-rail image.

[0068] Step 2: The wheel and rail profiles in the image are calculated by the frame difference method. Specifically, the grayscale of the corresponding pixels in two adjacent frames can be compared. If the grayscale of the corresponding pixels is very small when the ambient brightness does not change much, it can be considered that the object in this part is stationary. If the grayscale of a certain part of the image changes greatly, it can be considered that it is caused by the movement of the object in the image.

[0069] In this embodiment, rotational motion is not considered. The wheel remains relatively still with respect to the camera in the x, y, and z directions, and the background area of ​​the wheel is in motion relative to the camera. Therefore, the overall (or partial) contour of the wheel can be found by using the pixel grayscale difference between two adjacent frames of images. Similarly, the contour of the rail (in the image) can also be extracted.

[0070] like Figure 2 and Figure 3 As shown in the figure, 1 represents the laser light source camera, 2 represents the wheel, and 3 represents the rail. The positions of the wheel center point and the rail top in the image are calculated through the relationship between the geometric dimensions and the image coordinates. In the image coordinate system, assuming that P1 (x1, y1), P2 (x2, y2), and P3 (x3, y3) are any three points on the outer circumference of the wheel, according to the general equation of the circle:

[0071] x1 2 +y1 2 +Dx1+Ey1+F=0;

[0072] x2 2 +y2 2 +Dx2+Ey2+F=0;

[0073] x3 2 +y3 2 +Dx3+Ey3+F=0;

[0074] Then we can calculate the values ​​of D, E, and F, the coordinates of the center O of the circle (-D / 2, -F / 2). D and E represent translation parameters, which affect the horizontal and vertical coordinates of the center position of the circle. F represents a constant term, which affects the size of the circle (i.e., the radius).

[0075] Similarly, in the image coordinate system, assuming that G1(x4, y4) and G2(x5, y5) are any two points on the top of the rail, according to the general equation of a straight line:

[0076] Ax4+By4+C=0;

[0077] Ax5+By5+C=0;

[0078] The values ​​of A and B can be calculated, A represents the direction rate of the straight line equation of the rail top, B represents the inclination rate of the straight line equation of the rail top, and C represents the constant term, that is, the straight line equation of the rail top in the image coordinate system can be obtained, which can be used to approximately represent the position of the rail top in the image.

[0079] Step 3: In the image coordinate system, calculate the height from the center of the wheel to the top of the rail. Specifically, based on the wheel center point O(x0, y0), the straight line equation of the top of the rail Ax+By+C=0, the distance from the center of the wheel to the top of the rail can be calculated using the point-to-straight line distance formula.

[0080] Among them, the distance calculation formula is expressed as:

[0081]

[0082] Wherein, d represents the distance between the wheel center point and the rail top position, A represents the direction rate of the rail top straight line equation, B represents the inclination rate of the rail top straight line equation, C represents the constant term, x0 represents the horizontal coordinate of the wheel center point, and y0 represents the vertical coordinate of the wheel center point.

[0083] Step 4: Use camera calibration to obtain the conversion relationship between the image coordinate system and the world coordinate system. The camera calibration process uses the geometric relationship of imaging to calculate the internal and external parameters of the camera and find the projection relationship from the three-dimensional world coordinates to the two-dimensional image coordinates. The projection formula is expressed as:

[0084]

[0085] In the formula, λ represents the coordinate transformation coefficient, represents the coordinate point of the image coordinate system, E represents the internal parameters of the camera, R represents the rotation matrix, t represents the translation matrix, Represents the coordinate point in the world coordinate system.

[0086] The distance in the image coordinates can be converted to the distance in the world coordinates using λ: D = sd, where D represents the actual distance in the world coordinates and d represents the pixel distance in the image coordinates.

[0087] Step 5: Calculate the actual wheel lift based on the difference between the actual height and the theoretical height in step 4. Figure 4 and Figure 5 It is a schematic diagram of the wheel-rail contact state. The circle 70 mm away from the inner side of the rim is positioned as the rolling circle, the radius of the rolling circle is r, and the distance from the top of the rim to the rolling circle is defined as the rim height h;

[0088] Figure 4 Indicates that the wheel-rail is in normal position. Figure 5 It means that after the wheel climbs the rail, the wheel flange is at the top of the rail and there is a risk of derailment. When Dr ≥ h, it is judged that the wheel has climbed the rail and there is a risk of derailment, and an early warning and alarm can be issued based on the judgment result.

[0089] In summary, with the aid of the above technical solutions of the present invention, a vehicle wheel-rail state monitoring method proposed by the present invention analyzes images of wheels and wheel-rail contact areas based on monocular machine vision image technology, obtains the contour information of wheels and rails based on the analysis results, and detects the change in the height of the wheel center from the rail surface based on the contour information, thereby calculating the lift of the wheel relative to the rail, so that early warning and alarm can be issued after the train climbs the rails, derails, etc., thereby improving the safety of the train when in use. The present invention uses the frame difference method to perform gradient enhancement and filter enhancement processing on the wheel-rail image, analyzes the wheel contour and rail contour based on the pixel difference between the processed contour image and the initial image, facilitates the subsequent measurement of the rail surface lift, and lays the foundation for subsequent rail-wheel state monitoring.

[0090] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A wheel-rail condition monitoring method based on machine vision, characterized in that: The wheel-rail condition monitoring method comprises: The wheel-rail image of the wheel and rail area is obtained by using a light source camera, and the wheel and rail contour information in the wheel-rail image is extracted based on the frame difference method to determine the position of the wheel center point and the rail top; Analyze the distance between the center point of the wheel and the top of the rail, use the distance result and coordinate transformation relationship to analyze the actual wheel lift, and monitor the state of the wheel and rail based on the actual wheel lift to determine the risk of wheel derailment.

2. The wheel-rail condition monitoring method based on machine vision according to claim 1, characterized in that: The method of obtaining wheel-rail images of the wheel and rail area by using a light source camera, and extracting wheel and rail contour information in the wheel-rail images based on a frame difference method to determine the position of the wheel center point and the rail top includes: A laser light source camera installed at the end of the axle is used to project light to obtain a wheel-rail image covering the wheel and rail area; The frame difference method is used to perform filtering enhancement and channel feature extraction on the wheel-rail image, and a contour image is generated based on the extraction result. The contour image is then used to extract the wheel and rail contour information in the wheel-rail image. The equation information of the wheel and rail in the image coordinate system is analyzed based on the wheel and rail profile information, and the position of the wheel center point and the rail top is approximately analyzed based on the equation information results.

3. The wheel-rail condition monitoring method based on machine vision according to claim 2, characterized in that: The method of applying the frame difference method to filter and enhance the wheel-rail image and extract channel features, generating a contour image based on the extraction result, and extracting wheel and rail contour information in the wheel-rail image using the contour image includes: Applying the frame difference method to perform gradient enhancement processing on the wheel-rail image to generate a guide image, performing filter enhancement processing on the wheel-rail image based on the guide image and the filter, and outputting the initial image according to the processing result; Extract channel features from the primary image to obtain a secondary image, perform filtering and enhancement processing on the secondary image, and construct a pixel mean image based on the enhanced secondary image; A pixel value subtraction operation is performed on the secondary image and the pixel mean image to obtain a contour image, and the wheel contour and rail contour are analyzed based on the pixel difference between the contour image and the initial image.

4. The wheel-rail condition monitoring method based on machine vision according to claim 3 is characterized in that: The step of analyzing the equation information of the wheel and the rail in the image coordinate system based on the wheel and rail profile information, and approximately analyzing the position of the wheel center point and the rail top according to the equation information result includes: Generate a vehicle image coordinate system based on the wheel profile information, define three groups of wheel coordinate points outside the outer circumference of the wheel based on the vehicle image coordinate system, and obtain the coordinates of the wheel center point based on the wheel coordinate points and the circle equation information; Generate a rail image coordinate system based on the rail profile information, and define two groups of rail coordinate points on the rail top based on the rail image coordinate system; The straight line equation of the rail top in the image coordinate system is obtained according to the rail coordinate points and the straight line equation, and the straight line equation of the rail top is used to approximately represent the position of the rail top in the wheel-rail image.

5. The wheel-rail condition monitoring method based on machine vision according to claim 1, characterized in that: The analyzing the distance between the wheel center point and the rail top position, analyzing the actual wheel lift amount by using the distance result and the coordinate conversion relationship, and monitoring the wheel-rail state according to the actual wheel lift amount to determine the risk of wheel derailment includes: In the image coordinate system, a distance calculation formula is generated according to the straight line equation between the wheel center point and the rail top, and the distance between the wheel center point and the rail top position is calculated using the distance calculation formula; Determine the light source camera calibration process based on the camera's internal and external parameters, and use the calibration process results to find the coordinate projection relationship to obtain the coordinate conversion relationship between the image coordinate system and the world coordinate system; The distance result is converted into the distance in the world coordinate based on the coordinate conversion relationship to obtain the actual distance between the center point of the wheel and the top position of the rail, and the difference between the actual distance and the preset distance is calculated; The difference result is used as the actual wheel lift, and the state of the wheel-rail is judged based on the actual wheel lift. The risk of wheel derailment is monitored based on the wheel-rail state.

6. The wheel-rail condition monitoring method based on machine vision according to claim 5, characterized in that: The distance calculation formula is expressed as: Wherein, d represents the distance between the wheel center point and the rail top position, A represents the direction rate of the rail top straight line equation, B represents the inclination rate of the rail top straight line equation, C represents the constant term, x0 represents the horizontal coordinate of the wheel center point, and y0 represents the vertical coordinate of the wheel center point.

7. The wheel-rail condition monitoring method based on machine vision according to claim 6, characterized in that: The process of determining the light source camera calibration process based on the internal and external parameters of the camera, and using the calibration process result to find the coordinate projection relationship to obtain the coordinate conversion relationship between the image coordinate system and the world coordinate system includes: The pixel size, focal length and optical axis center of the light source camera are used as the camera internal parameters, and the rotation matrix and translation matrix are generated based on the geometric relationship of the light source camera imaging to calculate the camera external parameters; Convert the coordinate point from the world coordinate system to the camera coordinate system, obtain the coordinate of the coordinate point in the camera coordinate system, and convert the three-dimensional camera coordinate to the two-dimensional image plane; The projection formula is defined using the camera internal parameters to project the coordinate points in the two-dimensional image plane, and the projection relationship from the three-dimensional world coordinates to the two-dimensional image coordinates is analyzed based on the projection results; The camera calibration process is output according to the projection relationship, and the coordinate transformation relationship between the image coordinate system and the world coordinate system is obtained based on the camera calibration process.

8. The wheel-rail condition monitoring method based on machine vision according to claim 7, characterized in that: The expression of the projection formula is: In the formula, λ represents the coordinate transformation coefficient, represents the coordinate point of the image coordinate system, E represents the internal parameters of the camera, R represents the rotation matrix, t represents the translation matrix, Represents the coordinate point in the world coordinate system.

9. The wheel-rail condition monitoring method based on machine vision according to claim 8, characterized in that: The difference result is used as the actual wheel lift, and the wheel-rail state is determined according to the actual wheel lift. The risk of wheel derailment based on the wheel-rail state monitoring includes: The difference between the actual distance and the preset distance is used as the actual wheel lift, and the contact state between the wheel and the rail is determined based on the actual wheel lift; defining a position of a rolling circle and a radius of the rolling circle according to the contact state, and defining a wheel rim height based on a distance between the position of the rolling circle and a top position of the wheel; Obtain the difference between the actual distance and the rolling circle radius, compare the difference result with the wheel flange height to determine the state of the wheel and rail, and monitor the risk of wheel derailment based on the wheel and rail state.

10. The wheel-rail condition monitoring method based on machine vision according to claim 9, characterized in that: When the difference between the actual distance and the rolling circle radius is greater than or equal to the wheel flange height, it indicates that the wheel has climbed the rail and there is a risk of derailment.

Images