Tread base point detection method based on single-line structured light and application thereof

Abstract

The application provides a kind of tread base point detection method based on single line structured light, installs single line structured light sensor, train wheel shaft axis is not located on the light plane;Characteristic points and marks are provided on the calibration tread of the wheel;Sensor projects laser strip to the tread of the wheel, and the laser strip passes through the mark and is partially projected on the inner side of the wheel;The position of the mark on the light strip center line is recorded as point B;A straight line L is fitted using the points on the inner side of the wheel;The reference inner side and reference normal vector are obtained according to the straight line L and point B;When detecting, the sensor projects laser strip to the tread of the wheel;The light strip center line in the structured light image is obtained;The straight line is fitted using the points on the inner side of the wheel;The space plane passing through the straight line and the reference normal vector is recorded as the inner side of the wheel;The point with a distance of 70mm from the inner side of the wheel is recorded as the tread base point.This method does not need the light plane to pass through the train wheel shaft, reduces the installation and adjustment time of the sensor and the manufacturing cost of the sensor.

Description

Technical Field

[0001] This invention relates to the field of automated inspection in rail transit, specifically to a method for detecting tread base points based on single-line structured light and its application. Background Technology

[0002] In recent years, with the rapid development of my country's railways and the continuous increase in railway lines, the demand for intelligent inspection of train wheelsets has been growing. Among these, the compliance of wheel size parameters directly affects the safety of train operation, such as flange height, flange thickness, flange vertical wear, and QR value. These wheel dimensions need to be measured based on the location of the tread taping point. The tread taping point, a railway science and technology term published in 1997, refers to a point on the tread surface 70mm from the inner side of the wheel.

[0003] Currently, the mainstream methods for detecting tread base points are manual inspection or automatic inspection using vision sensors. Manual inspection is inefficient, has a high false detection rate, and is labor-intensive. Inspection using vision sensors usually requires fixing vision sensors around the wheel, emitting structured light from the vision sensors onto the wheel surface, and detecting wheel size parameters through the modulated structured light. Among them, visual sensors are divided into single-line structured light sensors and multi-line structured light sensors. Multi-line structured light sensors can emit multiple laser strips simultaneously, resulting in a larger measurement range in a single operation. However, they are expensive, emit low-brightness laser strips, and have lower light plane accuracy. In contrast, single-line structured light sensors are inexpensive, emit high-brightness laser strips, and have higher light plane accuracy. As the wheel rotates, emitting single-line structured light to different positions on the wheel can also achieve full detection of the wheel size. However, in the existing technology, when using single-line structured light sensors to measure wheel size, in order to meet the measurement accuracy requirements, it is necessary to ensure that the light plane projected by the sensor passes through the train wheel axle. For example, in the paper published by Feng Qibo et al.: Dynamic Measurement of Geometric Parameters and Defects of Train Wheelsets, this scheme is based on a one-dimensional laser displacement sensor (i.e., a single-line structured light sensor) to dynamically detect the geometric parameters of the wheel. In Section 2.2, the paper points out the technical problem that "the difficulty of the contour measurement method based on structured light lies in the fact that the laser plane projected by the laser does not coincide with the center of the measured wheelset." Therefore, existing solutions require ensuring that the light plane projected by the sensor passes through the train wheel axle, which increases the installation cost and time of the single-line structured light sensor. Furthermore, the sensor's position may shift after prolonged use, thus affecting the measurement accuracy. Summary of the Invention

[0004] To address the aforementioned technical issues, this paper proposes a wheel size detection method based on single-line structured light. Although this method uses a single-line structured light sensor for size detection, it eliminates the need for the light plane to pass through the train wheel axle, thus reducing sensor installation and debugging time as well as sensor manufacturing costs.

[0005] The technical solution is as follows:

[0006] A method for detecting tread reference points based on single-line structured light involves installing a single-line structured light sensor on the inner side of the wheel. The laser strip projected by the sensor can penetrate the wheel tread and cover a local area of ​​the inner side of the wheel. Meanwhile, the axis of the train wheel axle is not located on the light plane.

[0007] The reference normal vector of the inner surface of the wheel is obtained and stored through the following calibration process:

[0008] ①The standard wheel is a normal wheel without wear or a wheel mold that replicates a normal wheel without wear in proportion.

[0009] The tread surface of the calibration wheel is provided with one or more feature points along the circumference of the wheel, and the distance of each feature point from the inner side surface of the calibration wheel is a fixed length A; the position of each feature point is marked.

[0010] ② A single-line structured light sensor projects a laser strip onto the wheel tread, the laser strip passes through the marking and is partially projected onto the inner side of the wheel;

[0011] Acquire an image and extract the center line of the light stripe in the image; denote the position marked on the center line of the light stripe as point B;

[0012] ③ Fit a straight line L using the point on the center line of the light strip projected onto the inner side of the wheel;

[0013] The spatial plane that passes through the straight line L and is at a constant distance A from point B is denoted as the reference inner surface of the wheel, and the normal vector of the reference inner surface is stored as the reference normal vector.

[0014] During testing, the tread base point is obtained through the following steps:

[0015] 1) A single-line structured light sensor projects a laser strip that is partially projected onto the inner side of the wheel tread; an image is acquired, the image containing only the modulated laser strip;

[0016] 2) Obtain the center line of the light stripes in the structured light image;

[0017] The same method as the calibration process is used to determine the point on the center line of the light strip that is projected onto the inner side of the wheel, and a straight line is fitted using it; the spatial plane that passes through this straight line and whose normal vector is the reference normal vector is denoted as the inner side of the wheel.

[0018] 3) Mark the point 70mm away from the inner side of the wheel on the center line of the light strip as the tread base point.

[0019] Furthermore, in step ②, the acquired image contains only the modulated laser stripe;

[0020] The marking can block the laser strip, causing the laser strip marking to break at its location;

[0021] Alternatively, the mark can modulate the laser strip, causing the laser strip to deform at the position of the tread base point.

[0022] Preferably, the form of the mark is a black light-absorbing flat mark or a three-dimensional mark;

[0023] Step ② determines point B as follows:

[0024] When the marking is a black light-absorbing flat marking, the center line of the light strip breaks at the location of the marking, and the break point is point B.

[0025] When the sign is a three-dimensional sign, the center line of the light strip is deformed at the position of the sign, and the inflection point is point B.

[0026] Furthermore, in step ②, the images acquired are: a set of two-dimensional images and structured light images acquired at the same location;

[0027] The two-dimensional image contains only the calibrated wheel surface within the camera's field of view, or simultaneously contains the modulated laser stripe and the calibrated wheel surface within the camera's field of view; the structured light image contains only the modulated laser stripe.

[0028] The identifier can be displayed in a two-dimensional image.

[0029] Furthermore, in step ②, the identifier takes one of the following forms:

[0030] Form 1: A long strip-shaped marker, fixed to the tread surface with one side passing through each feature point;

[0031] Form 2: Dot-shaped markers, the number of which is the same as the number of tread base points, and each is fixed on a feature point;

[0032] Form 3: Rectangular raised markings, which are set on the tread surface and have a single edge passing through each feature point;

[0033] Form 4: Rectangular recessed markings, which are set on the tread surface and have a single edge passing through each feature point.

[0034] Furthermore, the method for determining point B is as follows:

[0035] First, determine the pixel coordinates of the feature points in the two-dimensional image based on the location of the markers; then, find the center point of the light stripe corresponding to the pixel coordinates in the structured light image.

[0036] Furthermore, if the two-dimensional image only contains the calibrated wheel surface within the camera's field of view, the method for determining the pixel coordinates of the tread base point based on the location of the marking is as follows:

[0037] When the label is elongated, the edge that passes through the feature point is recorded as the feature edge; multiple points are selected from the feature edge in the two-dimensional image and recorded as the pixel coordinates of the feature point;

[0038] When the marker is dot-shaped, the pixel at the geometric center of the marker is recorded as the pixel coordinate of the feature point in the two-dimensional image;

[0039] When the identifier is a rectangular raised or rectangular recessed identifier, the edge that passes through the feature point is recorded as the feature edge, and multiple points are selected from the feature edge in the two-dimensional image and recorded as the pixel coordinates of the feature point.

[0040] If a two-dimensional image simultaneously contains a modulated laser stripe and a calibrated wheel surface within the camera's field of view, the method for determining the pixel coordinates of feature points based on the location of the markings is as follows:

[0041] When the mark is long and narrow, the edge that passes through the base point of the tread is recorded as the feature edge. In the two-dimensional image, the intersection of the center line of the light strip and the feature edge is recorded as the pixel coordinate of the feature point.

[0042] When the marker is dot-shaped, in a two-dimensional image, the center point of the light stripe at the geometric center of the marker is recorded as the feature point;

[0043] When the marker is a rectangular raised marker or a rectangular recessed marker, the edge that passes through the feature point is recorded as the feature edge; in a two-dimensional image, the intersection of the light stripe center line and the feature edge is recorded as the feature point.

[0044] Preferably, in step ①, 2 to 10 feature points are provided;

[0045] In step ②, the wheel is rotated and the single-line structured light sensor projects laser strips onto each marker. Each time a laser strip is projected, an image is captured, and multiple points B are obtained.

[0046] In step ③, the center points of the light strips on the inner side of the wheel along multiple light strip center lines are used to fit a straight line L. Then, multiple points B and the straight line L are used to construct spatial plane equations. The optimal spatial plane is obtained by the optimization method and is denoted as the reference inner side.

[0047] Preferably, in step ③, the straight line L is fitted using the point on the center line of the light strip projected onto the inner side of the wheel, as follows:

[0048] The edge closest to the side of the wheel in the image is designated as the first edge; the 30-300 pixel points on the center line of the light strip closest to the first edge in the image are designated as points on the inner side of the wheel, and the straight line L is obtained by fitting it using the Ransac method.

[0049] In step ①, preferably, the feature point is the tread base point, which is measured by the fourth type of inspector;

[0050] In step 2), the method for obtaining the center line of the light stripe in the structured light image is the extremum method, the centroid method, or the Steger method.

[0051] The present invention also discloses a method for calculating the wheel flange geometry after obtaining the tread base point using the tread base point detection method described in claim 1, wherein the wheel flange geometry includes wheel flange height, wheel flange thickness, wheel flange vertical wear and QR value;

[0052] The method for calculating the flange height is as follows: calculate the vertical distance between each point in the flange area on the center line of the light strip and the tread base point, and record the maximum vertical distance as the flange height value;

[0053] The rim area is the area between the dividing point and the inner side of the wheel; the dividing point is the center point of the light strip, which is 30-50mm away from the tread base point.

[0054] The method for calculating the flange thickness is as follows: draw a straight line perpendicular to the inner side of the wheel through the tread base point, shift the line upward by 10mm or 12mm, and intersect it with the center line of the light strip. Record the horizontal distance between the two intersection points as the flange thickness value; record the intersection point closest to the inner side of the wheel as point I.

[0055] The method for calculating the vertical wear of the wheel flange is as follows: draw a straight line perpendicular to the inner side of the wheel through the tread base point, translate the line upward by 15mm and intersect it with the laser strip, and record the horizontal distance between the two points as the reference value; record the difference between the wheel flange thickness value and the reference value as the vertical wear value of the wheel flange.

[0056] The QR value is calculated as follows: the point corresponding to the wheel flange height value is recorded as the highest point. The highest point is lowered by 2mm to obtain point II. A straight line perpendicular to the inner side of the wheel is drawn through point II. This straight line intersects with the center line of the light strip. The intersection point away from the inner side of the wheel is recorded as point C. The horizontal distance between point I and point C is recorded as the QR value.

[0057] This method has the following advantages:

[0058] The reference normal vector is obtained and stored through a pre-calibration process, which does not occupy the actual inspection cycle and only needs to be performed once. During actual inspection, the reference normal vector is directly called to obtain the inner side of the wheel, the tread base point is obtained, and the geometric dimensions are obtained based on the tread base point to ensure the safe operation of the train.

[0059] In existing technologies, staff need to spend a lot of time and effort adjusting the installation position of the single-line structured light sensor so that the wheel axle is exactly on the light plane. However, this method does not require ensuring that the light plane passes through the train wheel axle, which reduces the installation requirements of the single-line structured light sensor. At the same time, the use of a single-line structured light sensor results in low equipment cost and high measurement accuracy. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of the wheelset structure in Example 1;

[0061] Figure 2 This is a schematic diagram showing the setting of strip markers at multiple feature points during the calibration process in Example 1;

[0062] Figure 3 This is a schematic diagram showing the setting of dot-shaped markers at multiple feature points during the calibration process in Example 1;

[0063] Figure 4 This is a schematic diagram showing the position of point B on the center line of the four light stripes in Example 1;

[0064] Figure 5 This is the grayscale image of the two-dimensional image acquired in step ② of Example 2;

[0065] Figure 6 This is a schematic diagram of calculating the rim geometry using the center line of the light stripe in Example 1. Detailed Implementation

[0066] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0067] Example 1

[0068] A method for detecting tread landmarks based on single-line structured light, such as Figure 1 As shown, a single-line structured light sensor is installed on the inside of the wheel. The laser strip it projects can penetrate the wheel tread and cover a part of the inner side of the wheel. At the same time, the axis of the train wheel axle is not located on the light plane.

[0069] The reference normal vector of the inner surface of the wheel is obtained and stored through the following calibration process:

[0070] ①The standard wheel is a normal wheel without wear or a wheel mold that replicates a normal wheel without wear in proportion.

[0071] One or more feature points are provided on the tread of the calibration wheel along the circumference of the wheel. The distance between each feature point and the inner side of the calibration wheel is a fixed length A. The positions of the feature points are marked.

[0072] The fixed length A can be obtained by measuring with calipers.

[0073] In a preferred embodiment, the feature point is the tread base point, which is measured by a fourth type of inspector;

[0074] ② A single-line structured light sensor projects a laser strip onto the wheel tread, which passes through the marking and has part of its projection on the inner side of the wheel;

[0075] Acquire an image and extract the center line of the light stripe in the image; denote the position marked on the center line of the light stripe as point B;

[0076] Among them, the methods for obtaining the center line of the light stripe in the structured light image are the extremum method, the centroid method, or the Steger method.

[0077] ③ Fit a straight line L using the point on the center line of the light strip projected onto the inner side of the wheel;

[0078] The spatial plane that passes through the straight line L and is at a constant distance A from point B is denoted as the reference inner surface of the wheel, and the normal vector of the reference inner surface is stored as the reference normal vector.

[0079] During testing, the tread base point is obtained through the following steps:

[0080] 1) A single-line structured light sensor projects a laser strip that is partially projected onto the inner side of the wheel tread; an image is acquired, which contains only the modulated laser strip; the installation position of the single-line structured light sensor is fixed, that is, its installation position remains consistent during the calibration process and the actual detection process.

[0081] 2) Obtain the center line of the light stripes in the structured light image;

[0082] The same method as the calibration process is used to determine the point on the center line of the light strip that is projected onto the inner side of the wheel, and a straight line is fitted using it; the spatial plane that passes through this straight line and whose normal vector is the reference normal vector is denoted as the inner side of the wheel.

[0083] Among them, the methods for obtaining the center line of the light stripe in the structured light image are the extremum method, the centroid method, or the Steger method.

[0084] 3) Mark the point 70mm away from the inner side of the wheel on the center line of the light strip as the tread base point.

[0085] In this embodiment, in step ②, the acquired image only contains the modulated laser stripe;

[0086] The marking can block the laser strip, causing the laser strip marking to break at its location;

[0087] Alternatively, the marker can modulate the laser strip, causing it to deform at the base point of the tread surface.

[0088] Specifically, the signage takes the form of a black, light-absorbing flat signage or a three-dimensional signage;

[0089] In detail, step ② determines point B as follows:

[0090] When the marking is a black light-absorbing flat marking, the center line of the light strip breaks at the location of the marking, and the break point is point B.

[0091] like Figure 2 As shown, black light-absorbing strip-shaped markers are set at each feature point. The area of ​​the strip-shaped markers can be set according to the actual situation; they can be narrow or wide.

[0092] Or, such as Figure 3 As shown, dot-shaped markers are set at each feature point;

[0093] When the sign is a three-dimensional sign (such as raised or recessed), the center line of the light strip is deformed at the position of the sign, and the inflection point is point B.

[0094] like Figure 4 As shown in the example, four feature points are set on the calibration wheel. The entire tread area to the right of the feature points is coated with black light-absorbing paint. A laser strip is projected onto the feature points to obtain the center lines of the four light strips (the light strip with a break in the middle indicates the area blocked by the wheel rim). The blank area on the right indicates the area where the light strip is broken due to being blocked by the black paint (the break point is point B).

[0095] In one preferred embodiment, step ① includes 2 to 10 feature points;

[0096] In step ②, the wheel is rotated and the single-line structured light sensor projects laser strips onto each marker. Each time a laser strip is projected, an image is captured, and multiple points B are obtained.

[0097] In step ③, the center points of the light strips on the inner side of the wheel along multiple light strip center lines are used to fit a straight line L. Then, multiple points B and the straight line L are used to construct spatial plane equations. The optimal spatial plane is obtained by the optimization method and is denoted as the reference inner side.

[0098] In comparison, the spatial plane obtained through optimization is more accurate.

[0099] More specifically, the straight line L is fitted using a point on the center line of the light strip projected onto the inner side of the wheel, as follows:

[0100] The straight line L is fitted using the point on the center line of the light strip projected onto the inner side of the wheel;

[0101] The edge closest to the side of the wheel in the image is designated as the first edge; the 30-300 pixel points on the center line of the light strip closest to the first edge in the image are designated as points on the inner side of the wheel, and the straight line L is obtained by fitting it using the Ransac method.

[0102] Application as a tread base point:

[0103] A method for calculating the flange geometry using the derived tread base points, the flange geometry including flange height, flange thickness, flange vertical wear and QR value;

[0104] Specifically, such as Figure 6 As shown, the method for calculating the flange height is as follows: calculate the vertical distance between each point in the flange area on the center line of the light strip and the tread base point, and record the maximum vertical distance as the flange height value;

[0105] The rim area is the area between the dividing point and the inner side of the wheel; the dividing point is the center point of the light strip, which is 30-50mm away from the tread base point.

[0106] The method for calculating the flange thickness is as follows: Draw a straight line perpendicular to the inner side of the wheel through the tread base point, shift the line upward by 10mm or 12mm, and intersect it with the center line of the light strip. Record the horizontal distance between the two intersection points as the flange thickness value; record the intersection point closest to the inner side of the wheel as point I.

[0107] The calculation method for vertical wear of the wheel flange is as follows: Draw a straight line perpendicular to the inner side of the wheel through the tread base point, translate the line upward by 15mm and intersect it with the laser strip, and record the horizontal distance between the two points as the reference value; record the difference between the wheel flange thickness value and the reference value as the vertical wear value of the wheel flange.

[0108] The QR value is calculated as follows: the point corresponding to the wheel flange height is recorded as the highest point. The highest point is lowered by 2mm to obtain point II. A straight line perpendicular to the inner side of the wheel is drawn through point II. This straight line intersects with the center line of the light strip. The intersection point away from the inner side of the wheel is recorded as point C. The horizontal distance between point I and point C is recorded as the QR value.

[0109] When there are multiple locations on the wheel that require the detection of tread base points, the wheel is rotated, and steps 1) to 3) are performed at each location to obtain multiple tread base points. Furthermore, the wheel flange geometry at each location to be measured can be calculated using these tread base points.

[0110] Example 2

[0111] The difference between this embodiment and Embodiment 1 is that:

[0112] In step ②, the images acquired are: a set of two-dimensional images and structured light images acquired at the same location;

[0113] Two-dimensional images contain only the surface of the calibrated wheel within the camera's field of view, or both the modulated laser stripe and the surface of the calibrated wheel within the camera's field of view; structured light images contain only the modulated laser stripe.

[0114] The logo can be displayed in a two-dimensional image.

[0115] In this embodiment, in step ②, the form of the identifier is one of the following:

[0116] Form 1: A long strip-shaped marker, fixed to the tread surface with one side passing through each feature point;

[0117] Form 2: Dot-shaped markers, the number of which is the same as the number of tread base points, and each is fixed on a feature point;

[0118] Form 3: Rectangular raised markings, which are set on the tread surface and have a single edge passing through each feature point;

[0119] Form 4: Rectangular recessed markings, which are set on the tread surface and have a single edge passing through each feature point.

[0120] In step ②, the method for determining point B is as follows:

[0121] First, determine the pixel coordinates of the feature points in the two-dimensional image based on the location of the markers; then, find the center point of the light stripe corresponding to the pixel coordinates in the structured light image.

[0122] Specifically, if the two-dimensional image only contains the calibrated wheel surface within the camera's field of view, the method for determining the pixel coordinates of the tread base point based on the location of the marking is as follows:

[0123] When the label is elongated, the edge that passes through the feature point is recorded as the feature edge; multiple points are selected from the feature edge in the two-dimensional image and recorded as the pixel coordinates of the feature point;

[0124] like Figure 5 As shown, the long, black strip-shaped marker obscures the entire tread area to the right of the feature point.

[0125] When the marker is dot-shaped, the pixel at the geometric center of the marker is recorded as the pixel coordinate of the feature point in the two-dimensional image;

[0126] When the identifier is a rectangular raised or rectangular recessed identifier, the edge that passes through the feature point is recorded as the feature edge, and multiple points are selected from the feature edge in the two-dimensional image and recorded as the pixel coordinates of the feature point.

[0127] If a two-dimensional image simultaneously contains a modulated laser stripe and a calibrated wheel surface within the camera's field of view, the method for determining the pixel coordinates of feature points based on the location of the markings is as follows:

[0128] When the mark is long and narrow, the edge that passes through the base point of the tread is recorded as the feature edge. In the two-dimensional image, the intersection of the center line of the light strip and the feature edge is recorded as the pixel coordinate of the feature point.

[0129] When the marker is dot-shaped, in a two-dimensional image, the center point of the light stripe at the geometric center of the marker is recorded as the feature point;

[0130] When the marker is a rectangular raised marker or a rectangular recessed marker, the edge that passes through the feature point is recorded as the feature edge; in a two-dimensional image, the intersection of the light stripe center line and the feature edge is recorded as the feature point.

[0131] The specific shape of the above-mentioned mark can be modified according to the actual situation. For example, the mark can be a polygon, a concentric circle, or a crosshair, as long as the feature point can be found.

[0132] The specific detection method for the tread base point in this embodiment is as follows:

[0133] A method for detecting tread reference points based on single-line structured light involves installing a single-line structured light sensor on the inner side of the wheel. The laser strip projected by the sensor can penetrate the wheel tread and cover a local area of ​​the inner side of the wheel. Meanwhile, the axis of the train wheel axle is not located on the light plane.

[0134] The reference normal vector of the inner surface of the wheel is obtained and stored through the following calibration process:

[0135] ①The standard wheel is a normal wheel without wear or a wheel mold that replicates a normal wheel without wear in proportion.

[0136] One or more feature points are provided on the tread of the calibration wheel along the circumference of the wheel. The distance between each feature point and the inner side of the calibration wheel is a fixed length A. The positions of the feature points are marked.

[0137] ② A single-line structured light sensor projects a laser strip onto the wheel tread, which passes through the marking and has part of its projection on the inner side of the wheel;

[0138] Acquire an image and extract the center line of the light stripe in the image; denote the position marked on the center line of the light stripe as point B;

[0139] ③ Fit a straight line L using the point on the center line of the light strip projected onto the inner side of the wheel;

[0140] The spatial plane that passes through the straight line L and is at a constant distance A from point B is denoted as the reference inner surface of the wheel, and the normal vector of the reference inner surface is stored as the reference normal vector.

[0141] During testing, the tread base point is obtained through the following steps:

[0142] 1) A single-line structured light sensor projects a laser strip that is partially projected onto the inner side of the wheel tread; an image is acquired, which contains only the modulated laser strip;

[0143] 2) Obtain the center line of the light stripes in the structured light image;

[0144] The same method as the calibration process is used to determine the point on the center line of the light strip that is projected onto the inner side of the wheel, and a straight line is fitted using it; the spatial plane that passes through this straight line and whose normal vector is the reference normal vector is denoted as the inner side of the wheel.

[0145] 3) Mark the point 70mm away from the inner side of the wheel on the center line of the light strip as the tread base point.

[0146] The method for calculating the wheel flange geometry using the derived tread base points is the same as in Example 1, and will not be repeated here.

[0147] The foregoing description of specific exemplary embodiments of the present invention is for illustrative and descriptive purposes. It is not intended to be exhaustive, nor to limit the invention to the precise forms disclosed; obviously, many changes and variations are possible in accordance with the foregoing teachings. The exemplary embodiments were chosen and described to explain the specific principles of the invention and its practical application, thereby enabling others skilled in the art to implement and utilize various exemplary embodiments of the invention, as well as their different alternatives and modifications. The scope of the invention is intended to be defined by the appended claims and their equivalents.

Claims

1. A method for detecting tread landmarks based on single-line structured light, characterized in that, A single-line structured light sensor is installed on the inside of the wheel. The laser strip it projects can penetrate the wheel tread and cover a part of the inner side of the wheel. At the same time, the axis of the train wheel axle is not located on the light plane. The reference normal vector of the inner surface of the wheel is obtained and stored through the following calibration process: ①The standard wheel is a normal wheel without wear or a wheel mold that replicates a normal wheel without wear in proportion. The tread surface of the calibration wheel is provided with one or more feature points along the circumference of the wheel, and the distance of each feature point from the inner side surface of the calibration wheel is a fixed length A; the position of each feature point is marked. ② A single-line structured light sensor projects a laser strip onto the wheel tread, the laser strip passes through the marking and is partially projected onto the inner side of the wheel; Acquire an image and extract the center line of the light stripe in the image; denote the position marked on the center line of the light stripe as point B; ③ Fit a straight line L using the point on the center line of the light strip projected onto the inner side of the wheel; The spatial plane that passes through the straight line L and is at a constant distance A from point B is denoted as the reference inner surface of the wheel, and the normal vector of the reference inner surface is stored as the reference normal vector. During testing, the tread base point is obtained through the following steps: 1) A single-line structured light sensor projects a laser strip that is partially projected onto the inner side of the wheel tread; an image is acquired, the image containing only the modulated laser strip; 2) Obtain the center line of the light stripes in the structured light image; The same method as the calibration process is used to determine the point on the center line of the light strip that is projected onto the inner side of the wheel, and a straight line is fitted using it; the spatial plane that passes through this straight line and whose normal vector is the reference normal vector is denoted as the inner side of the wheel. 3) Mark the point on the center line of the light strip that is 70mm away from the inner side of the wheel as the tread base point.

2. The tread surface reference point detection method based on single-line structured light as described in claim 1, characterized in that: In step ②, the acquired image contains only the modulated laser stripe; The marking can block the laser strip, causing the laser strip marking to break at its location; Alternatively, the mark can modulate the laser strip, causing the laser strip to deform at the position of the tread base point.

3. The tread surface reference point detection method based on single-line structured light as described in claim 1, characterized in that: The form of the logo is a black, light-absorbing flat logo or a three-dimensional logo; Step ② determines point B as follows: When the marking is a black light-absorbing flat marking, the center line of the light strip breaks at the location of the marking, and the break point is point B. When the sign is a three-dimensional sign, the center line of the light strip is deformed at the position of the sign, and the inflection point is point B.

4. The tread surface reference point detection method based on single-line structured light as described in claim 1, characterized in that: In step ②, the images acquired are: a set of two-dimensional images and structured light images acquired at the same location; The two-dimensional image contains only the calibrated wheel surface within the camera's field of view, or simultaneously contains the modulated laser stripe and the calibrated wheel surface within the camera's field of view; the structured light image contains only the modulated laser stripe. The identifier can be displayed in a two-dimensional image.

5. The tread surface reference point detection method based on single-line structured light as described in claim 4, characterized in that: In step ②, the identifier takes one of the following forms: Form 1: A long strip-shaped marker, fixed to the tread surface with one side passing through each feature point; Form 2: Dot-shaped markers, the number of which is the same as the number of tread base points, and each is fixed on a feature point; Form 3: Rectangular raised markings, which are set on the tread surface and have a single edge passing through each feature point; Form 4: Rectangular recessed markings, which are set on the tread surface and have a single edge passing through each feature point.

6. The tread surface reference point detection method based on single-line structured light as described in claim 5, characterized in that: The method for determining point B is as follows: First, determine the pixel coordinates of the feature points in the two-dimensional image based on the location of the markers; then, find the center point of the light stripe corresponding to the pixel coordinates in the structured light image.

7. The tread surface reference point detection method based on single-line structured light as described in claim 6, characterized in that: If the two-dimensional image only contains the surface of the calibrated wheel within the camera's field of view, the method for determining the pixel coordinates of the tread base point based on the location of the marking is as follows: When the label is elongated, the edge that passes through the feature point is recorded as the feature edge; multiple points are selected from the feature edge in the two-dimensional image and recorded as the pixel coordinates of the feature point; When the marker is dot-shaped, the pixel at the geometric center of the marker is recorded as the pixel coordinate of the feature point in the two-dimensional image; When the identifier is a rectangular raised or rectangular recessed identifier, the edge that passes through the feature point is recorded as the feature edge, and multiple points are selected from the feature edge in the two-dimensional image and recorded as the pixel coordinates of the feature point. If a two-dimensional image simultaneously contains a modulated laser stripe and a calibrated wheel surface within the camera's field of view, the method for determining the pixel coordinates of feature points based on the location of the markings is as follows: When the mark is long and narrow, the edge that passes through the base point of the tread is recorded as the feature edge. In the two-dimensional image, the intersection of the center line of the light strip and the feature edge is recorded as the pixel coordinate of the feature point. When the marker is dot-shaped, in a two-dimensional image, the center point of the light stripe at the geometric center of the marker is recorded as the feature point; When the marker is a rectangular raised marker or a rectangular recessed marker, the edge that passes through the feature point is recorded as the feature edge; in a two-dimensional image, the intersection of the light stripe center line and the feature edge is recorded as the feature point.

8. The tread surface reference point detection method based on single-line structured light as described in claim 1, characterized in that: In step ①, there are 2 to 10 feature points; In step ②, the wheel is rotated and the single-line structured light sensor projects laser strips onto each marker. Each time a laser strip is projected, an image is captured, and multiple points B are obtained. In step ③, the center points of the light strips on the inner side of the wheel along multiple light strip center lines are used to fit a straight line L. Then, multiple points B and the straight line L are used to construct spatial plane equations. The optimal spatial plane is obtained by the optimization method and is denoted as the reference inner side.

9. The tread surface reference point detection method based on single-line structured light as described in claim 1, characterized in that: Step ③: Fit a straight line L using the point on the center line of the light strip projected onto the inner side of the wheel, as follows: The edge closest to the side of the wheel in the image is designated as the first edge; the 30-300 pixel points on the center line of the light strip closest to the first edge in the image are designated as points on the inner side of the wheel. The straight line L is obtained by fitting it using the Ransac method. In step ①, the feature point is the tread base point, which is measured by the fourth type of inspector; In step 2), the method for obtaining the center line of the light stripe in the structured light image is the extremum method, the centroid method, or the Steger method.

10. A method for calculating the flange geometry after obtaining the tread base point using the tread base point detection method of claim 1, wherein the flange geometry includes flange height, flange thickness, flange vertical wear, and QR value; characterized in that: The method for calculating the flange height is as follows: calculate the vertical distance between each point in the flange area on the center line of the light strip and the tread base point, and record the maximum vertical distance as the flange height value; The rim area is the area between the dividing point and the inner side of the wheel; the dividing point is the center point of the light strip, which is 30-50mm away from the tread base point. The method for calculating the flange thickness is as follows: draw a straight line perpendicular to the inner side of the wheel through the tread base point, shift the line upward by 10mm or 12mm, and intersect it with the center line of the light strip. Record the horizontal distance between the two intersection points as the flange thickness value; record the intersection point closest to the inner side of the wheel as point I. The method for calculating the vertical wear of the wheel flange is as follows: draw a straight line perpendicular to the inner side of the wheel through the tread base point, translate the line upward by 15mm and intersect it with the laser strip, and record the horizontal distance between the two points as the reference value; record the difference between the wheel flange thickness value and the reference value as the vertical wear value of the wheel flange. The QR value is calculated as follows: the point corresponding to the wheel flange height value is recorded as the highest point. The highest point is lowered by 2mm to obtain point II. A straight line perpendicular to the inner side of the wheel is drawn through point II. This straight line intersects with the center line of the light strip. The intersection point away from the inner side of the wheel is recorded as point C. The horizontal distance between point I and point C is recorded as the QR value.

Images