Cross guide rail wear state monitoring method based on machine vision

Abstract

The invention relates to the field of wear monitoring, in particular to a cross guide rail wear state monitoring method based on machine vision, which comprises the following steps: acquiring an initial image and a standard image of a cross guide rail, subtracting the standard image from the initial image to obtain an error image, performing binarization processing on the error image by using an Otsu threshold method to obtain a binary image, and calculating the wear state of the cross guide rail; taking an area formed by white pixel points in the binary image as an abnormal area; calculating a texture complex factor and a shape factor of the abnormal area, and normalizing the product of the texture complex factor and the shape factor to obtain a wear characteristic value; and extracting an edge line of the abnormal area, calculating a mean value of gradient amplitudes of pixel points of the edge line, and carrying out weighted summation on the wear characteristic value and the mean value of the gradient amplitudes of the pixel points of the edge line to obtain a wear index for judging whether the surface of the cross guide rail is worn or not. According to the invention, the accuracy of cross guide rail state monitoring is improved.

Description

Technical Field

[0001] The present invention relates to the field of wear monitoring, and in particular to a method for monitoring the wear state of a cross guide rail based on machine vision. Background Art

[0002] Cross guides, as high-precision linear motion units, are used in CNC machine tools, semiconductor manufacturing equipment, precision measuring instruments, and other fields. Over time, these rails can wear out, directly impacting the machining accuracy and stability of the equipment. Therefore, monitoring the wear of these rails is crucial to ensure the accuracy and stability of the equipment.

[0003] Chinese patent publication CN110458824B discloses a drill bit wear detection method that uses image recognition to capture key characteristics of drill bit wear. The method then analyzes and identifies information such as the drill body, joint, matrix, cutting teeth, nozzle, blades, bearings, shank, and cone. The method then compares the information to a pre-set database, outputs drill bit wear detection and evaluation results, and provides drill bit optimization and improvement recommendations. If detection and identification are unsuccessful, the method can perform self-learning training to expand the database and improve the accuracy of drill bit wear identification.

[0004] The Variational Autoencoder (VAE) anomaly detection algorithm trains the VAE model on a large number of healthy (i.e., wear-free and interference-free) guideway images, enabling the model to deeply learn and accurately reconstruct healthy surface patterns. During online monitoring, if the input image contains patterns the model has not learned, such as physical wear, the VAE will be unable to effectively reconstruct them, resulting in significant reconstruction errors in abnormal areas. By detecting these significant errors, preliminary detection of potential defects is possible.

[0005] However, due to the working characteristics of the cross guide rail, its surface is prone to oil stains, water marks, reflections and other phenomena. In the process of using the VAE model to monitor the cross guide rail, its surface is easily disturbed by oil stains, water marks and reflections, and it is easy to mistakenly identify oil stains, water marks and reflections as wear defects, resulting in low accuracy of the monitoring results, which is not conducive to monitoring the wear status of the cross guide rail. Summary of the Invention

[0006] In order to solve the problem that the interference of the external environment on the cross guide rail leads to low accuracy of the monitoring results, the present invention provides a cross guide rail wear status monitoring method based on machine vision.

[0007] The present invention provides a method for monitoring the wear status of a cross guide rail based on machine vision, which adopts the following technical solutions: Obtain the initial image and standard image of the cross guide rail, subtract the standard image from the initial image to obtain an error map, binarize the error map using the Otsu threshold method to obtain a binary image, and define the area consisting of white pixels in the binary image as the abnormal area; calculate the texture complexity factor and shape factor of the abnormal area, and normalize the product of the texture complexity factor and shape factor to obtain the wear characteristic value; The edge lines of the abnormal area are extracted, the mean of the gradient amplitude of the edge line pixels is calculated, and the wear index is obtained by weighted summing of the wear characteristic value and the mean of the gradient amplitude of the edge line pixels to determine whether the cross guide rail surface is worn.

[0008] It can efficiently identify abnormal areas on the guide rail surface and form a wear index by fusing the wear characteristic value with the edge gradient amplitude, thereby accurately identifying the actual wear, reducing the interference of the external environment on the detection process, and effectively improving the accuracy of cross-guide rail status monitoring.

[0009] Preferably, the method for obtaining the standard image is: obtaining a historical image set of the cross guide rail, the historical image set containing multiple historical images of the cross guide rail; setting a corresponding label for each historical image, adding the initial image to the historical image set, clustering the labels in the historical image set to obtain multiple clusters, and obtaining the cluster cluster to which the initial image belongs; in the cluster cluster to which the initial image belongs, aligning the areas belonging to the cross guide rails in multiple historical images, sorting the pixels at the same position in the multiple historical images from small to large according to the grayscale value, calculating the average of the grayscale values ​​of the middle part, and using the average as the standard grayscale value of the position to further obtain the standard image.

[0010] By clustering historical images to generate standard images, we ensure that the standard images have good matching and representativeness under environmental variables such as light intensity and vibration frequency, thereby enhancing the adaptability and robustness under complex working conditions.

[0011] Preferably, the method for calculating the texture complexity factor of the abnormal area is as follows: mapping the abnormal area to a grayscale image, using the Canny operator to calculate the gradient direction of each pixel in the abnormal area, constructing multiple direction areas, each direction area has the same angle, and calculating the number and frequency of the gradient directions of the pixels in the abnormal area falling into each direction area. The expression of the texture complexity factor of the abnormal area is: ; Where H represents the texture complexity factor of the abnormal area, represents the frequency of the kth directional region, k represents the ordinal index of the directional region, and n represents the total number of directional regions.

[0012] By calculating the texture complexity factor based on the pixel gradient direction, the texture distribution differences in the image area can be accurately quantified, which is suitable for distinguishing real wear from pseudo-wear with a single texture, and improving the discriminative ability of anomaly detection.

[0013] Preferably, before calculating the shape factor of the abnormal area, the method further includes: calculating the straightness of the abnormal area, and the calculation method is: in the binary image, a rectangular coordinate system is constructed with any point as the origin, one coordinate axis is the x-axis, and the other coordinate axis is the y-axis, the position coordinates of each pixel point in the abnormal area are obtained, the position coordinates of each pixel point are used to construct a coordinate set, the coordinate set is used to construct a covariance matrix, and the eigenvalues ​​of the covariance matrix are calculated to further obtain the straightness of the abnormal area. The expression is: ; Where ZX represents the straightness of the abnormal area, 、 represent the maximum and minimum eigenvalues ​​respectively.

[0014] The covariance matrix is ​​used to construct the straightness index, and the wear area morphology is quantitatively analyzed in combination with the image geometric features, which can accurately identify typical linear scratch wear and improve the recognition accuracy.

[0015] Preferably, before calculating the shape factor of the abnormal area, the method further includes: calculating the roundness of the abnormal area. The calculation method is: in the binary image, extracting the edge line of the abnormal area, and calculating the perimeter and area of ​​the abnormal area, taking the number of pixels in the abnormal area as the area of ​​the abnormal area, and calculating the roundness of the abnormal area. The expression is: ; Where YD represents the roundness of the abnormal area, S represents the area of ​​the abnormal area, and L represents the circumference of the abnormal area.

[0016] Preferably, the expression of the shape factor is:

[0017] Where, represents the shape factor of the abnormal area, ZX represents the straightness of the abnormal area, YD represents the roundness of the abnormal area, and max represents the maximum value function.

[0018] By integrating straightness and roundness and establishing a comprehensive morphological evaluation index, the geometric structure of abnormal areas can be judged more comprehensively, which helps to further screen out irregularly shaped pseudo-wear areas.

[0019] Preferably, the wear index is calculated as follows: mapping the abnormal area to the grayscale image, extending the edge line outward by multiple pixels to obtain an auxiliary area, using the Otsu threshold method to perform binary classification on the gradient amplitude to obtain two clusters, calculating the mean of the gradient amplitude within the clusters, and taking the cluster with the larger mean as the first cluster. The expression of the wear index is:

[0020] Where, Indicates the wear index of the abnormal area, Represents the preset weight coefficient, represents the mean value of the gradient amplitude within the first cluster, Indicates the wear characteristic value of the abnormal area.

[0021] The introduction of auxiliary region and edge gradient clustering mechanism can enhance the significance of real wear edge features in the image, and combine them with texture morphological features to calculate the wear index. While retaining the real wear area, it effectively reduces the risk of non-wear areas such as transparent droplets being misidentified.

[0022] Preferably, the weight coefficient is calculated by calculating the intra-class variance and inter-class variance of the two clusters. The expression of the weight coefficient is:

[0023] Where, Represents the preset weight coefficient, represents the between-class variance of the two clusters, represents the intra-cluster variance of the two clusters, and tanh represents the hyperbolic tangent function.

[0024] Adaptive adjustment is performed based on the intra-class variance and inter-class variance, so that the wear index can dynamically balance the contribution of texture and edge features according to the current image characteristics, further improving the generalization ability and calculation accuracy of the model.

[0025] Preferably, the tag includes the light intensity and the vibration frequency of the cross guide.

[0026] Preferably, the labels in the historical image set are clustered using the DBSCAN clustering algorithm to obtain multiple clusters.

[0027] The DBSCAN clustering method is used to perform cluster analysis on image labels, which avoids dependence on the number of clusters, has stronger adaptability and practicality, and ensures the effectiveness of the standard image construction process.

[0028] The present invention has the following technical effects: The present invention can efficiently identify abnormal areas on the guide rail surface, and form a wear index by fusing the wear characteristic value with the edge gradient amplitude, thereby realizing accurate identification of real wear, reducing the interference of the external environment on the detection process, and being able to distinguish between real wear and false wear, thereby effectively improving the accuracy of cross-guide rail status monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1It is a flow chart of the cross guide rail wear status monitoring method based on machine vision of the present invention. DETAILED DESCRIPTION

[0030] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0031] The embodiment of the present invention discloses a method for monitoring the wear state of a cross guide rail based on machine vision, referring to Figure 1 , including the following steps, as follows: S1: Obtain an initial image of the cross rail, process the initial image to obtain a binary image, and further obtain the abnormal area.

[0032] A historical image set of the cross guide rail is obtained. The historical image set contains multiple historical images of the cross guide rail. The historical images are grayscale images. A corresponding label is set for each historical image. The label includes the light intensity and the vibration frequency of the cross guide rail. An industrial camera is used to collect the initial image and label of the cross guide rail, and the initial image is converted into a grayscale image. The initial image is added to the historical image set. The DBSCAN clustering algorithm is used to cluster the labels in the historical image set to obtain multiple clusters, and the cluster cluster to which the initial image belongs is further obtained. It can be understood that each cluster corresponds to multiple historical images, and the acquisition background (light intensity and vibration frequency) corresponding to the images in the same cluster is similar. Therefore, the grayscale values ​​corresponding to the images in the same cluster are also basically similar.

[0033] In the cluster to which the initial image belongs, there are multiple historical images corresponding to the cluster. The areas belonging to the cross guide rails in the multiple historical images are aligned, and the pixels at the same position in the multiple historical images are sorted from small to large according to the grayscale value. The mean of the grayscale value of the middle part is calculated, and the mean is used as the standard grayscale value of the position to further obtain the standard image. The standard image is subtracted from the initial image to obtain an error map. The error map is binarized using the Otsu threshold method to obtain a binary map. The binary map is processed into a connected domain to obtain the abnormal area (the area composed of white pixels).

[0034] In one embodiment, the grayscale values ​​in the middle portion specifically refer to: the sorted grayscale values ​​forming a grayscale value sequence, obtaining the median of the grayscale value sequence, and selecting multiple grayscale values ​​before and after the median as the grayscale values ​​in the middle portion.

[0035] In one embodiment, the grayscale value of the middle portion is obtained by forming a grayscale value sequence from the sorted grayscale values, obtaining the median of the grayscale value sequence, and selecting multiple grayscale values ​​before and after the median as the grayscale value of the middle portion. For example, the grayscale value of the 11th pixel in multiple historical images is sorted, and the 4th, 5th, 6th, 7th, and 8th grayscale values ​​are selected as the grayscale value of the middle portion. In historical images, noise data values ​​typically exhibit abnormal increases and decreases. Therefore, by taking the mean of the grayscale values ​​of the middle portion as the grayscale value of the standard image, the error of the standard image is reduced.

[0036] S2: Calculate the texture complexity factor of the abnormal area.

[0037] Based on the working environment of the cross guide rail, common real wear is usually scratches and pits, while pseudo wear is usually oil and water stains. The interior of the real wear is usually characterized by rich texture details, that is, it has high complexity, while pseudo wear is usually liquid spreading, and its texture details are relatively simple. Therefore, it is necessary to calculate the texture complexity factor of the abnormal area.

[0038] The calculation method is as follows: the abnormal area is mapped to the grayscale image. In the grayscale image, the Canny operator is used to calculate the gradient direction of each pixel in the abnormal area, and multiple direction areas are constructed. The angle of each direction area is the same. For example, since the spatial angle is 360°, the spatial angle is divided into 8 direction areas, each direction area is 45°, then the first direction area is (0-45°], the second direction area is (45-90°], ..., the eighth direction area is (270-360°], and the number and frequency of the gradient directions of the pixels in the abnormal area falling into each direction area are calculated. For example, there are 100 pixels in the abnormal area, of which the gradient directions of 20 pixels are located in the first direction area (0-45°), then the frequency of the first direction area is 20 / 100=0.2, and the expression of the texture complexity factor of the abnormal area is: ; Where H represents the texture complexity factor of the abnormal area, represents the frequency of the kth directional region, k represents the ordinal index of the directional region, and n represents the total number of directional regions. A larger value of the texture complexity factor indicates more texture details within the abnormal region, making it more likely to be true wear. Conversely, a smaller value of the texture complexity factor indicates fewer texture details within the abnormal region, making it more likely to be false wear.

[0039] S3: Calculate the shape factor of the abnormal area.

[0040] Since real wear is usually scratches and pits, scratches appear as approximately straight-line structures, while pits usually appear as approximately circular structures. The pseudo-wear areas formed by oil stains and water stains usually appear as irregular shapes. For example, oil stains and water stains form a structure with a large head and thin feet under the influence of gravity, or form a complex of straight lines and circles under the influence of external forces. Therefore, it is necessary to calculate the shape factor of the abnormal area to further distinguish between real wear and pseudo-wear.

[0041] S31: Calculate the straightness of the abnormal area.

[0042] The calculation method is as follows: in the binary image, a rectangular coordinate system is constructed with any point as the origin, where one coordinate axis is the x-axis and the other coordinate axis is the y-axis. The position coordinates of each pixel in the abnormal area are obtained, and the coordinates of each pixel are used to construct a coordinate set. The covariance matrix is ​​constructed using the coordinate set. The expression is:

[0043] Where, Indicates abnormal area The covariance matrix of represents the variance of the coordinate set in the x direction, represents the covariance of the coordinate set in the x-direction and the y-direction, Represents the variance of the coordinate set in the y direction. The covariance matrix is ​​symmetric, so equal .

[0044] Calculate the eigenvalues ​​of the covariance matrix and further obtain the straightness of the abnormal area. The expression is:

[0045] Where ZX represents the straightness of the abnormal area, 、 The eigenvalues ​​of the covariance matrix can reflect the degree of dispersion of the coordinate set in all directions. For a long and narrow linear area, the variance of its pixel coordinates along the major axis is greater than that along the minor axis. Therefore, when the shape of the abnormal area tends to be linear, its pixel distribution is highly concentrated in one direction, resulting in Much greater than , The value of approaches 0, and the value of straightness approaches 1; on the contrary, when the shape of the region is irregular, and The values ​​of are similar. The value of approaches 1, and the value of straightness approaches 0.

[0046] S32: Calculate the circularity of the abnormal area.

[0047] The calculation method is as follows: in the binary image, extract the edge line of the abnormal area and calculate the perimeter and area of ​​the abnormal area. Here, the number of pixels in the edge line is used as the perimeter of the abnormal area, and the number of pixels in the abnormal area is used as the area of ​​the abnormal area. Then, the roundness of the abnormal area is calculated. The expression is:

[0048] Where YD represents the roundness of the abnormal region, S represents the area of ​​the abnormal region, and L represents the perimeter of the abnormal region. When the shape of the abnormal region is close to a circle, the roundness value of the abnormal region approaches 1. When the shape of the abnormal region is irregular, its value approaches 0. Therefore, the roundness of the abnormal region can be used to determine the degree to which the shape of the abnormal region is close to a circle.

[0049] S33: Calculate the shape factor of the abnormal area using the straightness and abnormality of the abnormal area.

[0050] The expression for the shape factor is:

[0051] Where, represents the shape factor of the abnormal area, ZX represents the straightness of the abnormal area, YD represents the roundness of the abnormal area, and max represents the maximum value function.

[0052] On the cross guide rail, the shape of the wear area is usually manifested as linear scratch wear and pit wear formed by material shedding. The straightness of linear scratch wear approaches 1, and its roundness approaches 0. The roundness of pit wear approaches 1, and the straightness approaches 0. For the pseudo wear area formed by droplets and oil stains, due to its irregular shape, the value of its shape factor is small. Therefore, for an abnormal area, the maximum value of the straightness and roundness is taken to represent the shape factor of the abnormal area. It can be understood that the closer the value of the shape factor is to 1, the greater the possibility that it is real wear.

[0053] S4: Calculate the wear characteristic value of the abnormal area.

[0054] The expression is:

[0055] Where T represents the wear characteristic value of the abnormal area, represents the shape factor of the abnormal area, H represents the texture complexity factor of the abnormal area, and tanh represents the hyperbolic tangent function, which is used to map the value to (0, 1).

[0056] If the abnormal area is a real wear area, its shape factor and texture complexity factor will be larger, so the wear characteristic value obtained will be larger. The wear characteristic value obtained by calculation can further determine the possibility that the abnormal area is real wear.

[0057] S5: Calculate the wear index of the abnormal area using the wear characteristic value.

[0058] Cross guide rails operate in a complex environment. When inspecting the surface of a cross guide rail for wear, the presence of small impurity droplets, such as water or oil, often results in a larger calculated wear characteristic value. Therefore, using this wear characteristic value to determine wear can lead to misjudgments. Real wear areas have distinct edges, where the pixel gradients are large. Droplets are typically transparent and lack distinct edges, so the gradients at these edges don't change significantly. Therefore, we calculate the wear index for abnormal areas based on this characteristic to reduce misjudgments.

[0059] The abnormal area is mapped to the grayscale image, the edge line of the abnormal area is extracted, and the edge line is extended outward by a certain distance to obtain the auxiliary area. For example, 5 pixels are expanded outward; the gradient amplitude of each pixel in the auxiliary area is calculated using the Canny operator, and the gradient amplitude is normalized using the linear normalization method. Then, the gradient amplitude is binary classified using the Otsu threshold method to obtain two clusters. The mean of the gradient amplitude within the cluster is calculated, and the cluster with the larger mean is taken as the first cluster, and the cluster with the smaller mean is taken as the second cluster. The expression of the wear index is:

[0060] Where, Indicates the wear index of the abnormal area, Represents the preset weight coefficient, represents the mean value of the gradient amplitude within the first cluster, Indicates the wear characteristic value of the abnormal area.

[0061] The weight coefficient is calculated as follows: Calculate the inter-class variance of the first cluster and the second cluster , the weighted sum of the variance of the data points in the first cluster and the variance of the data points in the second cluster is taken as the intra-class variance ;

[0062]

[0063]

[0064] Where, Represents the preset weight coefficient, represents the between-class variance of the two clusters, represents the intra-class variance of the two clusters, represents the mean of the data points in the i-th cluster, Represents the mean value of the gradient amplitude of the pixels in the auxiliary area, Indicates the number of data points in the i-th cluster, i represents the index of the cluster, tanh represents the hyperbolic tangent function, which is used for normalization, and N represents the total number of data points in the two clusters. Represents the variance of the data points within the i-th cluster.

[0065] In the real wear area, the gradient amplitude of the edge pixel is larger. Therefore, the pixel corresponding to the data point in the first cluster is the edge pixel, and the pixel corresponding to the data point in the second cluster is the non-edge pixel. When calculating the wear index, if The larger the value of , the more obvious the difference between edge pixels and non-edge pixels is. In this case, the weight of the gradient amplitude should be increased. On the contrary, if The value of is small, indicating that the distinguishing features between edge pixels and non-edge pixels are not obvious enough. In this case, the weight of the gradient amplitude should be reduced and the weight of the wear feature value should be increased to improve the accuracy of the wear index calculation results.

[0066] S6: When the wear index is greater than the preset warning threshold, a warning prompt of cross guide rail wear is issued.

[0067] When the wear index exceeds the preset warning threshold, it indicates that the corresponding cross rail surface has real wear areas, and a warning prompt should be issued to facilitate repair or replacement of the cross rail. When the wear index is less than the preset warning threshold, the abnormal area on the cross rail surface is a pseudo-wear area and no additional treatment is required. The warning threshold is manually set according to actual conditions, and for example, the warning threshold is 0.8.

[0068] The above are all preferred embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A cross guide rail wear condition monitoring method based on machine vision, characterized in that: Including steps: Obtain the initial image and standard image of the cross guide rail, subtract the standard image from the initial image to obtain an error map, use the Otsu threshold method to binarize the error map to obtain a binary image, and regard the area composed of white pixels in the binary image as the abnormal area; Calculate the texture complexity factor and shape factor of the abnormal area, and normalize the product of the texture complexity factor and the shape factor to obtain the wear characteristic value; The edge lines of the abnormal area are extracted, the mean of the gradient amplitude of the edge line pixels is calculated, and the wear index is obtained by weighted summing of the wear characteristic value and the mean of the gradient amplitude of the edge line pixels to determine whether the cross guide rail surface is worn.

2. The cross guide rail wear state monitoring method based on machine vision according to claim 1, characterized in that: The method for obtaining the standard image is as follows: obtaining a historical image set of the cross guide rail, the historical image set containing multiple historical images of the cross guide rail; setting a corresponding label for each historical image, adding the initial image to the historical image set, clustering the labels in the historical image set to obtain multiple clusters, and obtaining the cluster to which the initial image belongs; In the cluster to which the initial image belongs, the areas belonging to the cross guide rails in multiple historical images are aligned, the pixels at the same position in multiple historical images are sorted from small to large according to the grayscale value, the mean of the grayscale value of the middle part is calculated, and the mean is used as the standard grayscale value of the position to further obtain the standard image.

3. The cross guide rail wear state monitoring method based on machine vision according to claim 1, characterized in that: The method for calculating the texture complexity factor of the abnormal area is as follows: map the abnormal area to the grayscale image, use the Canny operator to calculate the gradient direction of each pixel in the abnormal area, construct multiple direction areas, and each direction area has the same angle. Calculate the number and frequency of the gradient direction of the pixel points in the abnormal area falling into each direction area. The expression of the texture complexity factor of the abnormal area is: ; Where H represents the texture complexity factor of the abnormal area, represents the frequency of the kth directional region, k represents the ordinal index of the directional region, and n represents the total number of directional regions.

4. The method for monitoring the wear state of a cross guide rail based on machine vision according to claim 1, characterized in that: Before calculating the shape factor of the abnormal area, the following steps are also included: calculating the straightness of the abnormal area. The calculation method is: in the binary image, a rectangular coordinate system is constructed with any point as the origin, one coordinate axis is the x-axis, and the other coordinate axis is the y-axis. The position coordinates of each pixel in the abnormal area are obtained, and the position coordinates of each pixel are used to construct a coordinate set. The coordinate set is used to construct a covariance matrix, and the eigenvalues ​​of the covariance matrix are calculated. The straightness of the abnormal area is further obtained. The expression is: ; Where ZX represents the straightness of the abnormal area, 、 represent the maximum and minimum eigenvalues ​​respectively.

5. The cross guide rail wear state monitoring method based on machine vision according to claim 4 is characterized in that: Before calculating the shape factor of the abnormal area, the process also includes calculating the roundness of the abnormal area. The calculation method is as follows: in the binary image, the edge line of the abnormal area is extracted, and the perimeter and area of ​​the abnormal area are calculated. The number of pixels in the abnormal area is used as the area of ​​the abnormal area, and the roundness of the abnormal area is calculated. The expression is: ; Where YD represents the roundness of the abnormal area, S represents the area of ​​the abnormal area, and L represents the circumference of the abnormal area.

6. The method for monitoring the wear state of a cross guide rail based on machine vision according to claim 5, characterized in that: The expression for the shape factor is: Where, represents the shape factor of the abnormal area, ZX represents the straightness of the abnormal area, YD represents the roundness of the abnormal area, and max represents the maximum value function.

7. The cross guide rail wear state monitoring method based on machine vision according to claim 1, characterized in that: The calculation method of the wear index is as follows: the abnormal area is mapped to the grayscale image, the edge line is extended outward by multiple pixels to obtain the auxiliary area, the gradient amplitude is binary classified using the Otsu threshold method to obtain two clusters, the mean of the gradient amplitude within the cluster is calculated, and the cluster with the larger mean is taken as the first cluster. The expression of the wear index is: Where, Indicates the wear index of the abnormal area, Represents the preset weight coefficient, represents the mean value of the gradient amplitude within the first cluster, Indicates the wear characteristic value of the abnormal area.

8. The cross guide rail wear state monitoring method based on machine vision according to claim 7, characterized in that: The calculation method of the weight coefficient is: calculate the intra-class variance and inter-class variance of the two clusters. The expression of the weight coefficient is: Where, Represents the preset weight coefficient, represents the between-class variance of the two clusters, represents the intra-cluster variance of the two clusters, and tanh represents the hyperbolic tangent function.

9. The cross guide rail wear state monitoring method based on machine vision according to claim 2, characterized in that: Tags include light intensity and the vibration frequency of the cross-rails.

10. The cross guide rail wear state monitoring method based on machine vision according to claim 2, characterized in that: The DBSCAN clustering algorithm is used to cluster the labels in the historical image set to obtain multiple clusters.

Images