A method for measuring the creeping of railway turnout point rail

Abstract

The invention relates to a railway turnout point rail creeping measurement method, comprising the following steps: an image acquisition device is installed on a turnout sleeper, the shooting center line of the device is perpendicular to the vertical surface of the track, and images of the point rail and the base rail are acquired; preprocessing the image; edge extraction is performed on the image; a calibrated pixel ruler is made; pixel coordinates of the edge of the point rail are obtained; the actual distance between the edge of the point rail and the left boundary of an area of ​​interest is obtained by comparison with the calibrated pixel ruler, the distance is compared with an initial distance when the point rail does not creep, and the creeping direction and the creeping distance are calculated.

Description

Technical Field

[0001] The invention belongs to the field of intelligent operation and maintenance of rail transit, and relates to a railway turnout point rail creeping measurement method. Background Art

[0002] In the field of trunk railways and urban rail transit, turnout tip creep is also called turnout movement, which means that the turnout tip has a certain displacement relative to the basic rail, exceeding the standard value, and is a common turnout defect. According to the "Railway Line Repair Rules", turnout tip creep refers to the entire set of points being misaligned relative to the ground laying position, and moving longitudinally in the vertical direction. The tip creep of a seamless turnout shall not exceed 10mm. Turnout tip creep is an objective existence, and it is also a defect that is difficult to solve through maintenance.

[0003] At present, the detection of turnout point rail creeping diseases still relies on manual regular inspections and manual records and reports, which cannot grasp the parameters of turnout point rail creeping in real time, and there is a possibility of missed inspections and wrong inspections. Due to poor timeliness, low inspection efficiency, and the influence of skylight points, manual inspections cannot detect faults in a timely manner, which brings serious safety hazards to rail vehicle operation. Summary of the invention

[0004] Based on this, in view of the deficiencies in the prior art, the present invention aims to provide a railway turnout point rail creep measurement method to replace manual inspection and allow relevant personnel to grasp the state of turnout point rail creep in real time.

[0005] In order to achieve the above-mentioned purpose, the solution of the present invention is: the image acquisition device is installed on the turnout sleeper, the shooting direction is parallel to the track plane, and the vertical elevation images of the point rail and the base rail are collected; the image is preprocessed; the edge of the image is extracted; a calibrated pixel ruler is made; the pixel coordinates of the edge of the point rail are obtained; the actual distance between the edge of the point rail and the left boundary of the area of ​​interest is obtained by comparing with the calibrated pixel ruler, and the distance is compared with the initial distance when the point rail does not creep, and the creeping direction and creeping distance are calculated.

[0006] Furthermore, the preprocessing includes:

[0007] S21, extracting the region of interest: in the original image P0, manually set the coordinates of the four vertices of the region of interest (rectangle) According to the coordinates, the area is cut out from the image P0 and recorded as P, and the coordinates are The pixel point is set as the coordinate origin; where the superscript T represents the region of interest, and the subscript i from 0 to 3 represents the upper left corner, upper right corner, lower right corner and lower left corner of the image respectively;

[0008] S22, grayscale the image P intercepted in step S21 to obtain a grayscale image P g, the specific calculation formula is:

[0009]

[0010] in, is the pixel matrix of the blue (B) channel of the image, and the same is true for G and R; P mn is the pixel value corresponding to the mth row and nth column in a channel of the image P; parameters λ1, λ2, and λ3 are the weight coefficients corresponding to each channel. The weight coefficients are determined manually so that the grayscale difference between the sharp track boundary and other areas after grayscale conversion is larger.

[0011] Furthermore, the edge extraction includes:

[0012] S31. Using the gradient operator, calculate the grayscale image P g The gradient map is obtained by polling each pixel to obtain the candidate edge map P. pb ;

[0013] S32, Computational Graph P pb The amplitude histogram is calculated and the cumulative histogram is calculated to obtain the maximum gradient value S max ; Assume the target gradient interval width is S step , its value range is [S max / 5,S max ], let the starting point and end point of the target edge concentration area be S b and S e , let histogram S step The slope of the interval is k t , solve for S step and S b The value of S step ∈[S max / 5,S max ] and S b ∈[S max -S step ,1], and k t Reach the maximum value;

[0014] S33, set threshold S t , the target edge concentration area [S b ,S e ] is divided into two categories, namely [S b ,S t ] and [S t ,S e ], and count the number of pixels p with gradient i in the two categories respectively i , the maximum gradients in the two categories are t1 and t2 respectively, solve St The value of S t ∈[S b ,S e ],and

[0015]

[0016] Reach the maximum value;

[0017] S34, according to S t Binarize the image and get the edge map P b .

[0018] Furthermore, the step of making a calibration pixel ruler includes:

[0019] S41, collecting the vertical elevation images of the point rail and the base rail when the point rail is not creeping, and obtaining the grayscale image P of the image according to the steps in S2. g0 ;

[0020] S42, set the threshold p0, grayscale image P g0 Binarization is performed. The threshold value p0 is used for binarization. The threshold value should set only the pixels where the ruler scale is located to 1 as much as possible, and set the rest of the pixels to 0;

[0021] S43, manually set pixel height h C , forming a horizontal calibration line at the height, perpendicular to each scale line of the steel ruler; polling the pixel points on the horizontal calibration line, the coordinates are Among them, the superscript 0 indicates the horizontal calibration line; when the subscript i takes the maximum, As described in S21 When polling, if the current pixel is 1 and the previous pixel is 0, the pixel is considered to correspond to a scale line and recorded to form a pixel coordinate array and the tick array

[0022]

[0023] Furthermore, the pixel coordinates of the edge of the point track include:

[0024] S51, polling the horizontal coordinate x of the pixel point, that is, When x=x0 in a certain polling, count the number of pixels with the horizontal coordinate x0 that are 1, and record it as T;

[0025] S52. According to the result of S51, select the horizontal coordinate x that makes T the largest. T , which is the coordinate of the edge of the pointed rail;

[0026] Furthermore, the crawling direction and crawling distance are calculated based on the pixel coordinate array obtained in step S43. and the tick array And S52 gets the horizontal coordinate x T , perform interpolation calculations, including:

[0027] S61, find the distance from x T The nearest left value x T0 and the right value x T0+1 , and in the scale array according to the subscripts of these two values Find the corresponding scale value d T0 and d T0+1 ;

[0028] S62, calculate the actual distance d between the edge of the point rail and the left boundary of the area of ​​interest according to the following formula T ;

[0029]

[0030] D T Subtract the initial distance d0 when the point rail does not creep to obtain the creeping distance and creeping direction of the point rail. A positive result indicates that the point rail creeps to the right, and a negative result indicates that the point rail creeps to the left. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is a schematic diagram of the installation position of the image acquisition device of the present invention;

[0032] Figure 2 It is a block diagram of the working principle of the present invention. DETAILED DESCRIPTION

[0033] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] To facilitate the understanding of those skilled in the art, the preprocessing is now explained with an embodiment: the image resolution is 3840*2160, and the four vertices of the region of interest (rectangle) are set, with the coordinates of the upper left corner, upper right corner, lower right corner and lower left corner being (500, 200), (3200, 200), (3200, 1800), (500, 1800) respectively; and the weight coefficients of λ1=0.23, λ2=0.45, and λ3=0.32 are used for grayscale conversion.

[0035] In order to facilitate the understanding of those skilled in the art, edge extraction is now described using an embodiment: using a gradient operator Calculate the grayscale image Pg The gradient map is obtained, and according to the gradient values ​​of the two adjacent pixels on the left and right of each pixel, it is determined whether the gradient value of the current pixel is the maximum value. If it is, it is judged as a candidate edge point; otherwise, it is a non-edge point, and the gradient here is set to 0. According to S32, S33, S34, the gradient Figure 2 Value.

[0036] In order to facilitate the understanding of those skilled in the art, an embodiment is now used to illustrate the production of a calibrated pixel ruler: the image to be calibrated is collected and preprocessed to obtain a grayscale image; the grayscale image is binarized using a threshold of 210, and the pixel points where the ruler scale lines are located in this embodiment can be set to 1, and the remaining pixels can be set to 0; the pixel height is set to 1000, and the pixel coordinate array of the scale and the actual value array of the scale are obtained by polling.

[0037] To facilitate understanding by those skilled in the art, an embodiment is now used to illustrate how to obtain the pixel coordinates of the edge of the sharp rail: the horizontal coordinate of the pixel point is polled from 0 to 2700. When the coordinate is 1200, the number of pixels on the horizontal coordinate that is 1 is the largest, and this coordinate is the coordinate of the edge of the sharp rail.

[0038] In order to facilitate the understanding of those skilled in the art, the calculation of the crawling direction and crawling distance is now described with an embodiment: In this embodiment, the pixel width of the region of interest is 2700, the maximum scale of the ruler is 140.0mm, and the pixel coordinate array of the scale in this embodiment and the actual value array of the scale are in a linear relationship. According to the pixel coordinates of the edge of the sharp rail 1200, the distance from the edge of the sharp rail to the left boundary can be obtained as 62.2mm. According to the initial state distance of 70.0mm, the difference between the two is 7.8mm, so the sharp rail crawls 7.8mm to the left.

Claims

1. A railway turnout point rail creep measurement method, characterized in that: The following steps are involved: S1. The shooting direction of the image acquisition device is parallel to the track plane, and the vertical elevation images of the pointed rail and the basic rail are collected; S2, preprocessing the image; S3, extracting the edge of the image; S4, when the pointed rail does not creep, place a ruler on the surface of the base rail and align its zero scale line with the left boundary of the area of ​​interest in the picture, collect the above picture and make a calibrated pixel ruler, specifically: according to the steps described in S2, obtain the grayscale image P of the collected image g0 ; Set the threshold p0 and convert the grayscale image P g0 Binarization is performed. The threshold should set only the pixels where the ruler scale is located to 1 as much as possible, and the rest of the pixels to 0; Set pixel height h C , forming a horizontal calibration line at the height, perpendicular to each scale line of the steel ruler; polling the pixel points on the horizontal calibration line, the coordinates are The superscript 0 indicates the horizontal calibration line. If the current pixel is 1 and the previous pixel is 0, the pixel is considered to correspond to a scale line, and the value is recorded to form a pixel coordinate array. and the tick array S5, obtaining the pixel coordinates of the edge of the pointed rail; S6. Compare the actual distance between the edge of the sharp rail and the left border of the area of ​​interest by comparing with the calibrated pixel ruler, compare the distance with the initial distance when the sharp rail does not creep, and calculate the creeping direction and creeping distance.

2. The railway turnout point rail creep measurement method according to claim 1, characterized in that: The step S2 is to pre-process the image; specifically, it includes the following steps: S21, extracting the region of interest: in the original image P0, manually set the coordinates of the four vertices of the region of interest (rectangle) According to the coordinates, the area is cut out from the image P0 and recorded as P, and the coordinates are The pixel point is set as the coordinate origin; where the superscript T represents the region of interest, and the subscript i from 0 to 3 represents the upper left corner, upper right corner, lower right corner and lower left corner of the image respectively; S22, grayscale the image P intercepted in step S21 to obtain a grayscale image P g , the specific calculation formula is: in, is the pixel matrix of the blue (B) channel of the image, and the same is true for G and R; P mn is the pixel value corresponding to the mth row and nth column in a channel of the image P; parameters λ1, λ2, and λ3 are the weight coefficients corresponding to each channel. The weight coefficients are determined manually to make the grayscale difference between the sharp track boundary and other areas greater after grayscale conversion.

3. The railway turnout point rail creep measurement method according to claim 1, characterized in that: The step S3, edge extraction of the image; specifically, using an adaptive gradient threshold edge method to extract the vertical edge of the image to obtain a vertical edge map; comprises the following steps: S31. Use the gradient operator to calculate the grayscale image P g The gradient map is obtained by polling each pixel to obtain the candidate edge map P. pb ; S32, Computational Graph P pb The amplitude histogram is calculated and the cumulative histogram is calculated to obtain the maximum gradient value S max ; Assume the target gradient interval width is S step , its value range is [S max / 5,S max ], let the starting point and end point of the target edge concentration area be S b and S e , let histogram S step The slope of the interval is k t , solve for S step and S b The value of S step ∈[S max / 5,S max ] and S b ∈[S max -S step ,1], and k t Reach the maximum value; S33, set threshold S t , the target edge concentration area [S b ,S e ] is divided into two categories, namely [S b ,S t ] and [S t ,S e ], and count the number of pixels p with gradient i in the two categories respectively i , the maximum gradients in the two categories are t1 and t2 respectively, solve S t The value of S t ∈[S b ,S e ],and Reach the maximum value; S34, according to S t Binarize the image and get the edge map P b .

4. The railway turnout point rail creep measurement method according to claim 1, characterized in that: The step S5 is to obtain the pixel coordinates of the edge of the pointed rail; specifically, according to the edge map P obtained in step S34 b , search for vertical edge lines from the edge map, record the longest edge line as the longest vertical edge, and record its pixel horizontal coordinates; including the following steps: S51, polling the horizontal coordinate x of the pixel point, that is, When x=x0 in a certain polling, count the number of pixels with a horizontal coordinate of x0 that are 1, and record it as T; S52. According to the result of S51, select the horizontal coordinate x that makes T the largest. T , which are the coordinates of the edge of the pointed rail.

5. The railway turnout point rail creep measurement method according to claim 1, characterized in that: In step S6, the actual distance between the edge of the sharp rail and the left boundary of the region of interest is obtained by comparing with the calibrated pixel ruler, and the distance is compared with the initial distance when the sharp rail does not crawl, and the crawling direction and crawling distance are calculated. Specifically, according to the pixel coordinate array obtained in step S4, and the tick array And S52 gets the horizontal coordinate x T , perform interpolation calculation; The following steps are involved: S61, find the distance from x T The nearest left value x T0 and the right value x T0+1 , and in the scale array according to the subscripts of these two values Find the corresponding scale value d T0 and d T0+1 ; S62, calculate the actual distance d between the edge of the point rail and the left boundary of the area of ​​interest according to the following formula T ; D T Subtract the initial distance d0 when the point rail does not creep to obtain the creeping distance and creeping direction of the point rail. A positive result indicates that the point rail creeps to the right, and a negative result indicates that the point rail creeps to the left.

Images