A machine vision-based rail corrugation light-section image measurement method and device

Abstract

The application discloses a kind of steel rail corrugation light section image measurement method and equipment based on machine vision, which comprises obtaining the image of laser vertical irradiation on the surface of steel rail;Extract laser line from the image;Sub-pixel points in the laser line region of the image are extracted;The same waveform graph as the laser line in the image is obtained;Waveform graph is judged.The application uses the projection image of laser on the top surface of steel rail to judge the degree of corrugation of steel rail, and is not affected by the shaking of carrier, with high detection accuracy and good stability.It can run on the track for a long time, detect for a long distance and a wide range, and greatly improve the detection efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to a rail detection method and device, in particular to a method and device for detecting rail corrugation. BACKGROUND

[0002] With the development and expansion of the city, the importance of urban rail transit is highlighted. While the public is satisfied with the convenience provided by urban rail transit, they also pay more attention to the safety of urban rail transit. In the safety system of urban rail transit, the quality and safety of the steel rail is a very important part of the system, and its safety even directly determines whether the urban rail transit can operate normally.

[0003] The safety evaluation of the quality of the steel rail includes the surface crack of the steel rail, the loosening of the steel rail fastener, the steel rail corrugation and the like. Among them, the steel rail corrugation has a very important influence on the safe operation of urban rail transit. The steel rail corrugation mainly refers to the uneven wear in the wave shape on the top surface of the steel rail, which is actually the wave-shaped crushing on the steel rail. The substantial harm of the steel rail corrugation is mainly that the unevenness of the rail surface will cause a larger impact force of the train on the entire track structure, and the parts of the track connection will also be subjected to a larger impact force, resulting in loosening of the rail tie fastener and bolt, affecting the safety of train travel. At the same time, the steel rail corrugation will also cause strong vibration and noise of the train wheel-rail system, which will have a great impact on the comfort of passengers riding the train.

[0004] The existing method for detecting the steel rail corrugation is mainly based on the structured light detection method. This method mainly obtains the geometric deviation of the track in the two-dimensional plane space under the dynamic load by sensing the change of the geometric position between the reference light segment and the reference light segment through the imaging sensor, and further detects the unevenness of the track through the deviation. The device mainly detects the unevenness of the track by being mounted on a carrier. However, the carrier will vibrate during movement. Thus, the change of the geometric position between the reference light segment and the reference light segment depends not only on the unevenness of the track surface, but also on the vibration of the carrier, which causes the detection accuracy of the method to decay very badly, affecting the normal use of the product. SUMMARY

[0005] Therefore, the present application provides a steel rail corrugation light section image measurement method and device based on machine vision, which improves the accuracy and stability of the detection of the steel rail corrugation.

[0006] To solve the above technical problems, the present application provides a steel rail corrugation light section image measurement method based on machine vision, comprising:

[0007] obtaining an image of laser vertical irradiation on the surface of the steel rail;

[0008] extracting a laser line from the image;

[0009] A new laser curve is formed by extending sub-pixel points according to the pixel point information of the extracted laser line region;

[0010] A waveform graph of the laser line is obtained from the new laser curve;

[0011] The waveform graph parameters are obtained to determine whether there is an out-of-limit grinding.

[0012] As an improvement, the extraction of the laser line from the image comprises:

[0013] The laser line position region range image is obtained by cropping the image;

[0014] The image is subjected to an erosion operation;

[0015] The filtered image is subjected to an erosion operation;

[0016] The laser line in the image is extracted.

[0017] As an improvement, the extension of the sub-pixel points is:

[0018] The formula:

[0019]

[0020] The sub-pixel points are extracted; wherein P is the pixel value size of the original image, x, y is the pixel coordinate position, n is the accuracy of the sub-pixel points, and 1≤n≤4.

[0021] As an improvement, the extraction of the waveform graph of the laser line from the new laser curve comprises:

[0022] The sub-pixel points are subjected to Fourier transform to obtain the frequency spectrum image of the sub-pixel point y coordinate;

[0023] The frequency domain maximum value point is found on the frequency spectrum image, and the Fourier inverse transform is performed on the frequency domain maximum value point to obtain the laser line waveform image.

[0024] As an improvement, the formula:

[0025]

[0026] The sub-pixel points are subjected to Fourier transform; wherein N is a power of 2;

[0027] The formula:

[0028]

[0029] The Fourier inverse transform is performed on the frequency domain maximum value point.

[0030] As an improvement, it is judged whether the amplitude and wavelength in the waveform diagram exceed the threshold value, and if yes, it is considered that there is an out-of-limit wave abrasion to give an alarm.

[0031] The application also provides a steel rail wave abrasion light section image measuring device based on machine vision, comprising:

[0032] A laser emitter is used to emit laser to the top surface of the steel rail to present a straight laser line on the top surface of the rail;

[0033] An image collecting device is used to collect the image of the straight laser line formed by the laser on the top surface of the steel rail;

[0034] A controller is used to extract the laser line in the image collected by the image collecting device and generate a waveform diagram identical to the laser line, and analyze whether the amplitude and wavelength in the waveform diagram exceed the threshold value.

[0035] As an improvement, the laser emitter is located directly above the steel rail and perpendicular to the steel rail; the laser emitter is 190-210mm away from the top surface of the steel rail; the laser emitted by the laser emitter is a linear laser with a width of <0.25mm and a linear bending arc height of <0.3mm; the image collecting device is located on the side surface of the steel rail and forms an angle of 8-12° with the horizontal plane; the object distance of the image collecting device is 280-290mm; the relative positions of the laser emitter and the image collecting device are fixed.

[0036] As an improvement, a trolley for carrying the laser emitter and the image collecting device is further included, and the trolley can run on the steel rail; an encoder for collecting the rotating speed of the wheels of the trolley is further included, the speed of the trolley is calculated according to the rotating speed of the wheels, and the exposure time of the image collecting device is set according to the speed of the trolley.

[0037] As an improvement, the speed calculation formula of the trolley is

[0038] wherein C is the number of times of encoder signal triggering, R is the radius of the wheels of the trolley, T is time, and V is the speed of the trolley;

[0039] The exposure time calculation formula of the image collecting device is

[0040] wherein T camera is the exposure time, P is the resolution of the image collecting device, and V is the speed of the trolley.

[0041] The application has the advantages that the application uses the projection image of laser on the top surface of the steel rail to judge the wave abrasion degree of the steel rail, is not affected by the shaking of the carrier, has high detection accuracy and good stability, can run on the track for a long time, and can detect in a long distance and a large range, thereby greatly improving the detection efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0042] Figure 1 Flow chart of the present application.

[0043] Figure 2 Side view of the laser emitter and image acquisition device in the present application.

[0044] Figure 3 Front view of the laser emitter and image acquisition device in the present application.

[0045] Marked in the figure: 1 laser emitter, 2 image acquisition device, 3 steel rail. DETAILED DESCRIPTION

[0046] In order for those skilled in the art to better understand the technical solutions of the present application, the present application will be further described in detail below in combination with specific embodiments.

[0047] As shown in Figure 1 , the present application provides a steel rail corrugation light section image measurement method based on machine vision, the steps of which include:

[0048] S1 acquiring an image of laser vertically irradiating on the surface of the steel rail;

[0049] S2 extracting laser lines from the image;

[0050] S3 expanding sub-pixel points according to the pixel point information of the extracted laser line region to form a new laser curve;

[0051] S4 obtaining a waveform diagram of the laser line from the new laser curve;

[0052] S5 obtaining waveform diagram parameters for judging whether there is an out-of-limit corrugation.

[0053] Specifically, step S2 further includes:

[0054] S21 obtaining a laser line position region range image by cropping the image;

[0055] S22 performing an erosion operation on the image to retain the laser line region; a structure element module is constructed to find similar regions to the element module in the original image and retain them. Through the formula

[0056]

[0057] Performing an erosion operation on the image, where A is the original image pixel set, and B is the element module. The length* width of the element module B is 5*1 pixel.

[0058] S23 filters the image after the corrosion; specifically, the image after the corrosion is filtered through expansion. The expansion operation of the image can be used to eliminate the small area of noise in the image, and can further smooth the laser image of the rail surface captured by the camera.

[0059] S24 extracts the laser line in the image. Specifically, the image binarization method is used to extract the laser line region of the filtered image.

[0060] Step S3 is specifically using the formula:

[0061]

[0062] The sub-pixel point is extracted; wherein P is the pixel value size of the original image, x, y is the pixel coordinate position, and n is the accuracy of the sub-pixel point. In this embodiment, 1≤n≤4. Due to the influence of the image acquisition device resolution and the ambient light, during the shooting process, the original image obtained by shooting will lose part of the real information. In order to restore these information, the concept of sub-pixel point is proposed. The sub-pixel point is based on the original image pixel point, and the lost real information is restored by using the original image gradient, brightness and other information, which is a supplement to the original image information, and is more accurate than the original pixel point.

[0063] In order to obtain the accurate amplitude and wavelength size of the corrugation, Fourier transform is needed for the obtained sub-pixel coordinates to obtain a sinusoidal wave most similar to the original waveform. By extracting the amplitude and wavelength of the sinusoidal wave, the amplitude and wavelength size of the corrugation can be obtained, and the harm can be further judged. Step S4 specifically includes:

[0064] S41 performs Fourier transform on the sub-pixel point to obtain the frequency spectrum image of the sub-pixel point y coordinate;

[0065] Using the formula:

[0066]

[0067] The sub-pixel point is subjected to Fourier transform; wherein N is a power of 2. Through fast Fourier transform, the frequency spectrum image of y value can be obtained.

[0068] S42 finds the frequency domain maximum value point on the frequency spectrum image, and performs inverse Fourier transform on the frequency domain maximum value point to obtain the laser line waveform image.

[0069] Using the formula:

[0070]

[0071] The frequency domain maximum value point is subjected to inverse Fourier transform.

[0072] On the basis of the spectrum image of the y value, in order to be able to fit the best waveform, the maximum value point in the frequency domain is found on the spectrum image, and the point is combined with the Fourier inverse transform formula to obtain the best waveform signal image. After the Fourier inverse transform formula, the waveform graph most similar to the original waveform can be obtained, and the amplitude and wavelength can be used to judge the original wave grinding condition to a certain extent.

[0073] In step S5, whether the amplitude and wavelength in the waveform graph exceed the threshold value is judged, and if so, it is considered that there is an out-of-limit wave grinding to alarm. For example, according to the pixel resolution size of the image, which is 0.02 mm, and combined with the wave grinding alarm requirement of the work inspection, which is that the distance between the wave peak and the wave trough is 0.5 mm. Therefore, if the pixel point difference between the wave peak and the wave trough of the sine wave in the waveform graph is 25, it can be judged that the amplitude of the sine wave reaches 0.02 mm*50=0.5 mm, that is, it is out of limit.

[0074] As shown in Figure 2 , Figure 3 The present application also provides a steel rail wave grinding light section image measuring device based on machine vision, which comprises: a laser emitter 1 for emitting laser to the top surface of a steel rail 3 to present a straight laser line on the top of the rail; an image acquisition device 2 for acquiring the straight laser line image formed by the laser on the top surface of the steel rail 1; and a controller for extracting the laser line in the image acquired by the image acquisition device 2 and generating a waveform graph same as the laser line, and analyzing whether the distance between the wave peak and the wave trough in the waveform graph exceeds a threshold value.

[0075] The laser emitter 1 is located directly above the steel rail 3 and perpendicular to the steel rail 3, and the distance from the top surface of the steel rail 3 is 190-210 mm. And the laser emitted by the laser emitter 1 is a linear laser, and the width is <0.25 mm, and the linear bending arc height is <0.3 mm.

[0076] The image acquisition device 2 is located on the side of the steel rail 3 and forms an angle of 8-12° with the horizontal plane, and the object distance is 280-290 mm.

[0077] The relative positions of the laser emitter 1 and the image acquisition device 2 are fixed. They can be fixed by a support. After the relative positions are fixed, the vibration of the carrier will not change the relative positions of the two, so it will not have a negative impact on the detection result.

[0078] In order to enable the laser emitter and the image acquisition device to travel along the steel rail, a trolley for carrying the laser emitter 1 and the image acquisition device 2 is further included, and the trolley can run on the steel rail. In addition, an encoder for acquiring the wheel speed of the trolley is further included, the speed of the trolley is calculated according to the wheel speed, and the exposure time of the image acquisition device 2 is set according to the speed of the trolley.

[0079] The speed calculation formula of the trolley is:

[0080] ,

[0081] Wherein C is the encoder signal trigger times, R is the radius of the trolley wheel, T is the time, V is the trolley speed.

[0082] The exposure time calculation formula of the image acquisition device 2 is:

[0083] ,

[0084] Wherein T camera is the exposure time, P is the resolution of the image acquisition device 2, and V is the trolley speed.

[0085] The above is only the preferred embodiment of the present application, it should be pointed out that the above preferred embodiment should not be considered as limiting the present application, the protection scope of the present application should be limited by the scope defined by the claims. For ordinary skilled in the art, without departing from the spirit and scope of the present application, a number of improvements and refinements can also be considered as the protection scope of the present application.

Claims

1. A machine vision-based method of measuring rail corrugation light-section images, characterized by It comprises: acquiring an image of laser vertical irradiation on the surface of a steel rail; extracting a laser line from the image; extending sub-pixel points according to pixel point information of the extracted laser line region to form a new laser curve; calculating a waveform graph of the laser line from the new laser curve, comprising: performing Fourier transform on the sub-pixel points to obtain a frequency spectrum image of the y coordinate of the sub-pixel points; finding a frequency domain maximum value point on the frequency spectrum image, and performing Fourier inverse transform on the frequency domain maximum value point to obtain a laser line waveform image; calculating waveform graph parameters for judging whether there is an over-limit wave abrasion.

2. The machine vision-based rail corrugation light-section imaging measurement method according to claim 1, characterized in that The extraction of the laser line from the image comprises: obtaining a laser line position region range image by cropping the image; performing an erosion operation on the image; performing filtering on the eroded image; extracting the laser line in the image.

3. The machine vision-based rail corrugation light-section imaging measurement method according to claim 1, characterized in that The extension of the sub-pixel points is: The sub-pixel point is extracted by using a formula wherein P is a pixel value size of the original image, x and y are pixel coordinate positions, and n is a precision of the sub-pixel point, 1≤n≤4.

4. The machine vision-based rail corrugation light-section imaging measurement method of claim 1, wherein judging whether the amplitude and wavelength in the waveform graph exceed a threshold value, and if so, considering that there is an over-limit wave abrasion to alarm.

5. A machine vision-based rail corrugation light-section image measuring apparatus for implementing the machine vision-based rail corrugation light-section image measuring method according to any one of claims 1 to 4, characterized by It comprises: a laser emitter for emitting laser to the top surface of a steel rail to present a straight laser line on the top of the rail; an image acquisition device for acquiring an image of the straight laser line formed by the laser on the top surface of the steel rail; a controller for extracting a laser line in the image acquired by the image acquisition device and generating a waveform graph identical to the laser line, and analyzing whether the amplitude and wavelength in the waveform graph exceed a threshold value.

6. A machine vision-based rail corrugation light-section imaging device according to claim 5, characterized in that: The laser emitter is located directly above the steel rail and is perpendicular to the steel rail; the distance between the laser emitter and the top surface of the steel rail is 190-210 mm; the laser emitted by the laser emitter is linear laser with a width of <0.25 mm and a linear bending arc height of <0.3 mm; the image acquisition device is located on the side of the steel rail and forms an angle of 8-12° with the horizontal plane; the object distance of the image acquisition device is 280-290 mm; the relative positions of the laser emitter and the image acquisition device are fixed.

7. A machine vision-based rail corrugation light-section imaging device according to claim 5, characterized in that: It further comprises a trolley for carrying the laser emitter and the image acquisition device, wherein the trolley can run on the steel rail; and further comprises an encoder for acquiring the rotation speed of the wheels of the trolley, calculating the speed of the trolley according to the rotation speed of the wheels, and setting the exposure time of the image acquisition device according to the speed of the trolley.

8. The machine vision-based steel rail wave abrasion light section image measuring device according to claim 7, characterized in that: the speed calculation formula of the trolley is where C is the number of encoder signal triggers, R is the radius of the trolley wheel, T is the time, and V is the trolley velocity; the exposure time calculation formula of the image acquisition device is where T camera is the exposure time, P is the image acquisition device resolution, and V is the trolley speed.

Images