CN112109769B - Wheel tread damage detection method based on urban rail vehicle wheel tread damage detection system - Google Patents
Wheel tread damage detection method based on urban rail vehicle wheel tread damage detection system Download PDFInfo
- Publication number
- CN112109769B CN112109769B CN202011040267.5A CN202011040267A CN112109769B CN 112109769 B CN112109769 B CN 112109769B CN 202011040267 A CN202011040267 A CN 202011040267A CN 112109769 B CN112109769 B CN 112109769B
- Authority
- CN
- China
- Prior art keywords
- wheel
- wheel tread
- tread
- image
- shooting
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61K—AUXILIARY EQUIPMENT SPECIALLY ADAPTED FOR RAILWAYS, NOT OTHERWISE PROVIDED FOR
- B61K9/00—Railway vehicle profile gauges; Detecting or indicating overheating of components; Apparatus on locomotives or cars to indicate bad track sections; General design of track recording vehicles
- B61K9/12—Measuring or surveying wheel-rims
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Studio Devices (AREA)
Abstract
The invention discloses a system and a method for detecting wheel tread damage of an urban rail vehicle, relates to a system and a method for detecting train faults, and aims to solve the problems that panoramic images of wheel treads obtained by the existing system for detecting wheel tread damage are often low in definition and cause more errors in damage identification, and the system comprises: the wheel sensor is used for generating a shooting trigger signal when a wheel passes by; the control module is used for generating a shooting command according to the shooting trigger signal; the image acquisition control and processing module is used for sending the shooting command to the camera module; the camera module is used for acquiring partial images of the wheel tread surface after receiving a shooting command, and acquiring a plurality of partial images of the wheel tread surface of the same wheel; the partial image of the wheel tread surface is sent to an image acquisition control and processing module; the image acquisition control and processing module is also used for splicing the multiple wheel tread surface partial images of the same wheel into a complete wheel tread surface after receiving the wheel tread surface partial images.
Description
Technical Field
The invention relates to a train fault detection system and method, in particular to a system and method for identifying damage by splicing complete images of wheel tread.
Background
Wheel tread damage is a common failure of wheel set treads in rail vehicles as a result of excessive braking forces during vehicle operation causing the wheels to skid on the rail surface. The resulting flats caused by the gouges cause severe impacts of the wheels with the rails, which impacts will seriously damage the bearing system and roadbed of the vehicle. The impact is a very dangerous defect and seriously threatens the driving safety.
Particularly, after the speed of the train is increased, the requirement on the precision of the shape geometric dimension of the wheel is higher. In emergency braking, the collision and relative sliding between the wheel set and the steel rail can cause tread scratching or peeling, and the peeling can seriously affect the safety and service life of trains and rail facilities.
With the increasing of the speed of the trains and the increase of the heavy-duty trains in China, the abrasion of the wheel sets is increasingly accelerated, the damage condition of the wheel treads can be detected by the existing wheel tread damage detection system without stopping the trains, but due to the fact that the speed of the trains is too high, the obtained wheel tread panoramic images are often low in definition, and more errors are caused during damage identification.
Disclosure of Invention
The invention aims to solve the problems that panoramic images of wheel treads obtained by the existing wheel tread damage detection system are often low in definition and have more errors during damage identification, and provides a wheel tread damage detection system and method for urban rail vehicles.
The invention discloses a system for detecting damage to wheel treads of urban rail vehicles, which comprises a wheel sensor, a control module, an image acquisition control and processing module and a camera module, wherein the wheel sensor is connected with the control module;
the wheel sensor is electrically connected with the control module and used for generating a shooting trigger signal when a wheel passes by and sending the shooting trigger signal to the control module;
the control module is electrically connected with the image acquisition control and processing module and used for generating a shooting command according to the shooting trigger signal and sending the shooting command to the image acquisition control and processing module;
the image acquisition control and processing module is electrically connected with the camera module and is used for sending a shooting command to the camera module;
the camera module is used for acquiring partial images of the wheel tread surface after receiving a shooting command, and acquiring a plurality of partial images of the wheel tread surface of the same wheel; the partial image of the wheel tread surface is sent to an image acquisition control and processing module;
the image acquisition control and processing module is also used for splicing the multiple wheel tread surface partial images of the same wheel into a complete wheel tread surface after receiving the wheel tread surface partial images.
The wheel tread damage detection method based on the urban rail vehicle wheel tread damage detection system comprises the following specific steps:
step one, image acquisition: each front wheel tread image shooting camera and each rear wheel tread image shooting camera shoot to obtain 1/4 wheel treads and 1/8 wheel treads, and the 4 front wheel tread image shooting cameras and the 4 rear wheel tread image shooting cameras on the same side send the shot wheel tread partial images of each wheel to the image acquisition control and processing module;
step two, image preprocessing: determining a partial image saliency region of the wheel tread, cutting and correcting the 1/4 wheel tread to obtain a 1/4 wheel tread in a rectangular form, and extracting a 1/4 wheel tread accurate region by using a 1/8 wheel tread by adopting an image projection method;
step three, segmenting partial image of the wheel tread: simultaneously operating multiple 1/4 wheel treads to obtain a segmented tread area, and then performing gray level transformation to ensure that the average gray level of the multiple 1/4 wheel treads are the same; then selecting each segmented tread area according to the actual splicing requirement;
step four, edge fitting: firstly, finding out a rim edge line of a segmented tread area image, then fitting edges, fitting bent edges into vertical edges, and removing the influence of tread radian of a wheel to obtain a fitted image;
step five, splicing each section of image: aiming at the fitted images, according to the image acquisition speed, utilizing corresponding characteristic points or edges in the overlapped public view field areas among the images to finish 1/4 wheel tread seamless splicing to obtain a complete wheel tread;
step six, roughly positioning damage of the wheel tread: obtaining a gray threshold, carrying out binarization on the complete wheel tread, and carrying out edge detection and morphological processing to obtain a damaged area position;
seventhly, detecting and identifying damage of the wheel tread: and realizing automatic identification of the tread damage category by adopting an image template matching and feature extraction algorithm.
The invention has the beneficial effects that:
the invention obtains the panoramic image of the wheel tread by the high-speed camera arranged on the side of the track. The damage of the wheel tread is reflected on the image in the view field range of the camera, and the wheel tread damage model is established through the image processing of the computer, so that the damage of the wheel tread can be detected comprehensively and accurately, and the specific effects are as follows:
1. 360-degree full-circle detection of the wheel pair can be realized;
2. by shooting 1/4 and 1/8 wheel tread images, the 1/4 wheel tread image is used for splicing 360-degree full-circle images, and the 1/8 wheel tread image assists in detecting the edge area of 1/4 wheel tread image, so that the detection precision is improved;
3. the damage condition of the wheel tread can be known more comprehensively and accurately, and the non-stop online real-time detection of the 360-degree full-circumference tread of the train wheel is realized through the coordinated work of the multiple cameras.
Drawings
FIG. 1 is a schematic structural diagram of a wheel tread damage detection system of an urban rail vehicle according to the invention;
FIG. 2 is a schematic diagram of a matching structure of a wheel sensor, a camera module and a rail in the system for detecting the damage to the wheel tread of the urban rail vehicle; the one-way arrow is the driving direction, L is the distance between the two far-end magnetic steels, and L is the distance between the two front wheel tread image shooting cameras or the rear wheel tread image shooting cameras;
FIG. 3 is a schematic view of the working principle of the system for detecting damage to the wheel tread of an urban rail vehicle according to the invention; wherein, the one-way arrow is the driving direction.
Detailed Description
In a first specific embodiment, the system for detecting damage to the wheel tread of the urban rail vehicle comprises a wheel sensor 1, a control module 2, an image acquisition control and processing module 3 and a camera module 4;
the wheel sensor 1 is electrically connected with the control module 2 and used for generating a shooting trigger signal when a wheel passes by and sending the shooting trigger signal to the control module 2;
the control module 2 is electrically connected with the image acquisition control and processing module 3 and is used for generating a shooting command according to the shooting trigger signal and sending the shooting command to the image acquisition control and processing module 3;
the image acquisition control and processing module 3 is electrically connected with the camera module 4 and is used for sending a shooting command to the camera module 4;
the camera module 4 is used for acquiring partial images of the wheel tread surface after receiving the shooting command, and acquiring a plurality of partial images of the wheel tread surface of the same wheel; and the partial image of the wheel tread is sent to the image acquisition control and processing module 3;
the image acquisition control and processing module 3 is further configured to splice the multiple wheel tread surface partial images of the same wheel into a complete wheel tread surface after receiving the wheel tread surface partial images.
Specifically, as shown in fig. 1, the system for detecting the damage to the wheel tread of the urban rail vehicle comprises a wheel sensor 1, a control module 2, an image acquisition control and processing module 3 and a camera module 4;
the wheel sensor 1 comprises far-end magnetic steel and a trigger module;
the control module 2 and the image acquisition control and processing module 3 are respectively a control industrial personal computer and an image acquisition industrial personal computer; the image industrial personal computer adopts a built-in kilomega network card, and one image industrial personal computer is provided with four network cards for collecting the wheel tread images shot by the camera module 4.
The camera module 4 is arranged on the outer side of the track; still include light source module, with the cooperation of camera module 4, carry out the light filling to detecting wheel tread region through the angle of design, and camera module 4 shoots the wheel tread region of light source module light filling.
And the control industrial personal computer transmits the number of the vehicle, the wheel position, the vehicle position and the sequential signal of the acquired image to the image acquisition industrial personal computer through the serial port. Meanwhile, door opening and closing and fan signals are sent out through the I/O board card, the image acquisition industrial personal computer starts to acquire wheel tread partial images after receiving a shooting command, the acquired wheel tread partial images are processed to form 360-degree full-circle wheel tread images of the wheel, and the defect states of scratching, peeling and the like of the wheel set tread for one circle can be visually and clearly observed on the remote control center computer.
Best embodiment, this embodiment is a further description of the first embodiment, and the camera module 4 includes a front wheel camera set 4-1 and a rear wheel camera set 4-2;
the front wheel shooting camera set 4-1 comprises a plurality of pairs of front wheel tread image shooting cameras 4-1-1, and each pair of front wheel tread image shooting cameras 4-1-1 are symmetrically distributed on the outer sides of the two tracks and are respectively used for collecting wheel tread partial images of front wheels on two sides;
the distances between every two front wheel tread image shooting cameras 4-1-1 are equal;
the rear wheel shooting camera group 4-2 comprises a plurality of pairs of rear wheel tread image shooting cameras 4-2-1, and each pair of rear wheel tread image shooting cameras 4-2-1 are symmetrically distributed on the outer sides of the two rails and are respectively used for collecting wheel tread partial images of rear wheels on two sides;
and the distances between each pair of rear wheel tread image shooting cameras 4-2-1 are equal.
Best embodiment, this embodiment is a further description of the first embodiment, and the wheel sensor 1 includes a front wheel trigger module group 1-1 and a rear wheel trigger module group 1-2;
the front wheel trigger module group 1-1 comprises front wheel trigger modules 1-1-1 the number of which is the same as that of the front wheel tread image shooting cameras 4-1-1, each front wheel trigger module 1-1 corresponds to each pair of front wheel tread image shooting cameras 4-1-1 one by one, and each front wheel trigger module 1-1-1 generates a shooting trigger signal corresponding to the front wheel tread image shooting camera 4-1-1 when a wheel passes by and sends the shooting trigger signal to the control module 2;
the rear wheel trigger module group 1-2 comprises rear wheel trigger modules 1-2-1 the number of which is the same as that of the rear wheel tread image shooting cameras 4-2-1, each rear wheel trigger module 1-2-1 corresponds to each pair of rear wheel tread image shooting cameras 4-2-1 one by one, and each rear wheel trigger module 1-2-1 generates a shooting trigger signal corresponding to the rear wheel tread image shooting camera 4-2-1 when a wheel passes by and sends the shooting trigger signal to the control module 2.
Specifically, the system aims at a wheel bogie braked by a brake shoe, and a front wheel middle brake shoe and a rear wheel middle brake shoe can shield a camera from shooting a tread design. According to the brake shoe braking condition of the metro vehicle bogie, the front wheels are suitable for shooting at the front part of the bogie, and the rear wheels are suitable for shooting at the rear part of the bogie;
the camera module 4 is composed of a front wheel shooting camera set 4-1 and a rear wheel shooting camera set 4-2, and each camera is provided with a light source to supplement light for the shooting range of the camera. The front wheel and the rear wheel on one side of each bogie are respectively provided with 4 front wheel tread image shooting cameras 4-1-1 and rear wheel tread image shooting cameras 4-2-1, the front wheel trigger module 1-1 and the rear wheel trigger module 1-2-1 are used for controlling the shooting of the wheel tread according to signals sent by the front wheel trigger module 1-1 and the rear wheel trigger module 1-2-1 in sequence, and the non-stop on-line real-time detection of the 360-degree full-circumference tread of the train wheel is realized through the coordination work of multiple cameras.
Best embodiment, this embodiment is a further description of the first embodiment, the number of the front wheel trigger modules 1-1-1 is m, and the number of pairs of the front wheel tread image capturing cameras 4-1-1 is m pairs; m pairs of front wheel tread image shooting cameras 4-1-1 are arranged at intervals of 1/m times of the wheel tread circumference;
each front wheel trigger module 1-1-1 is arranged at a position which can shoot 1/m times of wheel tread circumference in front of the corresponding front wheel tread image shooting camera 4-1-1; wherein m is an integer of 3 or more.
In a preferred embodiment, the present embodiment is further described with respect to the first embodiment, where the shooting command includes a front wheel instant shooting command and a front wheel delayed shooting command;
the lens direction of the front wheel tread image shooting camera 4-1-1 faces the rear of the running of the vehicle, and when each front wheel tread image shooting camera 4-1-1 receives a corresponding front wheel instant shooting command, the collected partial image of the wheel tread is 1/m of the wheel tread; when a corresponding front wheel delayed shooting command is received, the collected wheel tread partial image is 1/2m of the wheel tread;
and the m front wheel tread image shooting cameras 4-1-1 on the same side shoot sequentially along with the advancing of the vehicle, and at least m continuous wheel tread partial images are collected.
Best embodiment, this embodiment is a further description of the first embodiment, the number of the rear wheel trigger modules 1-2-1 is n, and the number of pairs of the rear wheel tread image capturing cameras 4-2-1 is n; the distance between the multiple pairs of rear wheel tread image shooting cameras 4-2-1 is 1/n of the wheel tread circumference;
each rear wheel trigger module 1-2-1 is arranged at a position behind the corresponding rear wheel tread image shooting camera 4-2-1, and can shoot 1/2n wheel tread circumferences; wherein n is an integer of 3 or more.
In a preferred embodiment, the present embodiment is further described with respect to the first embodiment, where the shooting command further includes a rear wheel instant shooting command and a rear wheel delayed shooting command;
the orientation of a lens of the rear wheel tread image shooting camera 4-2-1 is the same as the advancing direction of the vehicle, and when each rear wheel tread image shooting camera 4-2-1 receives a corresponding rear wheel delay shooting command, the collected partial image of the wheel tread is 1/n of the wheel tread; when a corresponding rear wheel instant shooting command is received, the collected wheel tread partial image is 1/2n of the wheel tread;
and the n rear wheel tread image shooting cameras 4-2-1 at the same side shoot sequentially along with the advancing of the vehicle and at least collect n continuous wheel tread partial images.
The front wheel shooting camera set 4-1 and the rear wheel shooting camera set 4-2 are different in installation angle, the shooting direction of the front wheel shooting camera set 4-1 is over against the advancing direction of the vehicle, and the shooting direction of the rear wheel shooting camera set 4-2 is the same as the advancing direction of the vehicle.
The cameras in the front wheel shooting camera group 4-1 and the rear wheel shooting camera group 4-2 are distributed at equal intervals in the track direction according to the wheel circumferences, and each camera shoots a wheel tread larger than 90 degrees.
Best embodiment, this embodiment is a further description of the first embodiment, the number of the front wheel trigger modules 1-1-1 is 4, and the number of pairs of the front wheel tread image capturing cameras 4-1-1 is 4;
each front wheel trigger module 1-1-1 is arranged at the circumference of the 1/4 wheel tread in front of the corresponding front wheel tread image shooting camera 4-1-1;
the partial image of the wheel tread collected by each front wheel tread image shooting camera 4-1-1 is 1/4 wheel tread image and 1/8 wheel tread image; 4 front wheel tread image shooting cameras 4-1-1 on the same side shoot images in sequence along with the advancing of the vehicle, and at least 4 continuous wheel tread partial images are collected;
the number of the rear wheel trigger modules 1-2-1 is 4, and the number of pairs of rear wheel tread image shooting cameras 4-2-1 is 4;
each rear wheel trigger module 1-2-1 is arranged at the circumference of the 1/8 wheel tread behind the corresponding rear wheel tread image shooting camera 4-2-1;
the partial image of the wheel tread collected by each rear wheel tread image shooting camera 4-2-1 is 1/4 wheel tread image and 1/8 wheel tread image; and 4 rear wheel tread image shooting cameras 4-2-1 on the same side shoot in sequence along with the advancing of the vehicle, and at least 4 continuous wheel tread partial images are collected.
The camera module 4 consists of a front wheel shooting camera set 4-1 and a rear wheel shooting camera set 4-2, 16 cameras are needed in total, and each camera is provided with a light source module to supplement light for the shooting range of the camera.
The front wheel shooting camera set 4-1 and the rear wheel shooting camera set 4-2 are totally four image acquisition industrial personal computers for carrying out image acquisition on 16 cameras.
As shown in FIG. 2, the front wheels on both sides of the vehicle are sequentially photographed by four front wheel tread image photographing cameras 4-1-1 and R5, R6, R7, R8 from R1, R2, R3, R4, respectively, and the wheel tread surface partial images from the four front wheel tread image photographing cameras 4-1-1; the partial images of the wheel tread surfaces shot by the R1, R2, R3 and R4 are collected by a first image collection industrial personal computer, and the partial images of the wheel tread surfaces shot by the R5, R6, R7 and R8 are collected by a second image collection industrial personal computer.
The rear wheels on two sides of the vehicle are sequentially shot by four rear wheel tread image shooting cameras 4-2-1 of R9, R10, R11 and R12 and four rear wheel tread image shooting cameras 4-2-1 of R13, R14, R15 and R16 respectively; the partial images of the wheel tread surfaces shot by the R9, the R10, the R11 and the R12 are collected by a third image collection industrial personal computer, and the partial images of the wheel tread surfaces shot by the R13, the R14, the R15 and the R16 are collected by a fourth image collection industrial personal computer.
The front wheel trigger module 1-1-1 corresponding to the front wheel camera set 4-1 is arranged at the position of the shooting tread 1/4 in front of each front wheel camera set 4-1. When the industrial personal computer is controlled to receive a signal of one front wheel trigger module 1-1-1, triggering a corresponding front wheel tread image shooting camera 4-1-1 to shoot through software, simultaneously supplementing light to the front wheel tread image shooting camera 4-1-1 through a light source module, shooting 1/4 wheel tread images of wheels through the front wheel tread image shooting camera 4-1-1, controlling the industrial personal computer to shoot 1/8 wheel tread images of the wheels through speed delay and triggering the same front wheel tread image shooting camera 4-1-1 for the second time; and sequentially shooting by the subsequent front wheel tread image shooting camera 4-1-1 according to the shooting mode to obtain 1/4 wheel tread images and 1/8 wheel tread images of the front wheel pair.
As shown in fig. 3, for example, when the front wheel is shot, the upper image of the wheel may have a lower detection accuracy of the edge region of the wheel tread image because the information of the 1/4 wheel tread image is shot but the upper portion occupies less pixel points of the wheel tread image because of the shooting angle of the front wheel tread image shooting camera 4-1-1 and the shape of the wheel; and the edge area of 1/4 wheel tread surface images can be repeatedly shot by additionally shooting 1/8 wheel tread surface images, so that the accuracy of image recognition is improved.
The rear wheel trigger module 1-2-1 corresponding to the rear wheel shooting camera group 4-2 is arranged at the position of the shooting tread 1/8 behind each rear wheel tread image shooting camera 4-2-1. When the industrial personal computer is controlled to receive a signal of one rear wheel trigger module 1-2-1, triggering a corresponding rear wheel tread image shooting camera 4-2-1 to shoot through software, simultaneously supplementing light to the rear wheel tread image shooting camera through a light source module, shooting 1/8 wheel tread images of wheels through the camera, controlling the industrial personal computer to shoot 1/4 wheel tread images of the wheels through speed delay and triggering the same rear wheel tread image shooting camera 4-2-1 for the second time; and sequentially shooting by the subsequent rear wheel tread image shooting camera 4-2-1 according to the shooting mode to obtain 1/4 wheel tread images and 1/8 wheel tread images of the rear wheel pair.
Best embodiment, this embodiment is a further description of the first embodiment, and the wheel sensor 1 further includes distal end magnetic steels 1-3;
the far-end magnetic steel 1-3 is electrically connected with the control module 2 and is used for sending a vehicle receiving signal to the control module 2 when a vehicle passes by; the vehicle-sending device is also used for sending a vehicle-sending signal to the control module 2 when the vehicle arrives and leaves;
the control module 2 is also electrically connected with the door opening and closing device of the rail side chassis and is used for opening the door opening and closing device of the rail side chassis after receiving the vehicle receiving signal; the door opening and closing device is also used for closing the rail-side chassis after receiving the car sending signal;
the camera module 4 is arranged in the rail-side chassis.
Specifically, a camera and a light source in the camera module 4 are both placed in a camera box, the camera box is sealed, a fan is installed on the side of the camera box, and dustproof and waterproof glass is installed at the lens of the camera and the light emitting position of the light source; the camera box is installed in the rail-side chassis. The door opening and closing motor drives the door opening and closing device, when a vehicle arrives and the industrial personal computer is controlled to receive a signal of the far-end magnetic steel, the door opening and closing device is opened, when the front wheel trigger module 1-1-1 and the rear wheel trigger module 1-2-1 are triggered by wheels in the period, the front wheel shooting camera set 4-1 and the rear wheel shooting camera set 4-2 shoot, and after the vehicle leaves, the door opening and closing device is closed.
In a second specific embodiment, the wheel tread damage detection method based on the wheel tread damage detection system for the urban rail vehicle in the second embodiment specifically includes the following steps:
step one, image acquisition: each front wheel tread image shooting camera 4-1-1 and each rear wheel tread image shooting camera 4-2-1 shoot to obtain 1/4 wheel treads and 1/8 wheel treads, the 4 front wheel tread image shooting cameras 4-1-1 and the 4 rear wheel tread image shooting cameras 4-2-1 on the same side send the shot wheel tread partial images of each wheel to the image acquisition control and processing module 3;
step two, image preprocessing: determining a partial image saliency region of the wheel tread, cutting and correcting the 1/4 wheel tread to obtain a 1/4 wheel tread in a rectangular form, and extracting a 1/4 wheel tread accurate region by using a 1/8 wheel tread by adopting an image projection method;
step three, segmenting partial image of the wheel tread: simultaneously operating multiple 1/4 wheel treads to obtain a segmented tread area, and then performing gray level transformation to ensure that the average gray level of the multiple 1/4 wheel treads are the same; then selecting each segmented tread area according to the actual splicing requirement;
step four, edge fitting: firstly, finding out a rim edge line of a segmented tread area image, then fitting edges, fitting bent edges into vertical edges, and removing the influence of tread radian of a wheel to obtain a fitted image;
step five, splicing each section of image: aiming at the fitted images, according to the image acquisition speed, utilizing corresponding characteristic points or edges in the overlapped public view field areas among the images to finish 1/4 wheel tread seamless splicing to obtain a complete wheel tread;
step six, roughly positioning damage of the wheel tread: obtaining a gray threshold, carrying out binarization on the complete wheel tread, and carrying out edge detection and morphological processing to obtain a damaged area position;
seventhly, detecting and identifying damage of the wheel tread: and realizing automatic identification of the tread damage category by adopting an image template matching and feature extraction algorithm.
Specifically, the urban rail vehicle wheel tread damage detection system provided by the invention adopts a camera to shoot tread images, and based on machine vision and a mode recognition principle, an image processing and mode recognition technology is applied to realize an integrated solution scheme of dynamic acquisition of wheel set images, extraction of wheel image tread regions, generation of wheel tread whole-circle images, positioning of tread damage regions, damage degree detection and the like, so that the aim of dynamic detection of urban rail vehicle wheel tread damage is fulfilled.
1. Image acquisition: the cameras and the light source receive the external trigger signal to complete the tread image acquisition of the wheels, each camera takes two tread images, and the 4 cameras send the acquired tread images taken by each wheel to the image acquisition industrial personal computer. And the image acquisition industrial personal computer processes the acquired wheel tread image.
2. Image preprocessing, namely determining a significant region of a tread image and cutting the image; then, the cut image is corrected into a rectangular tread image; and (4) finishing the extraction of the accurate region of the tread image by adopting an image projection method.
3. Roughly positioning damage of the wheel tread, solving a gray threshold value by adopting an Otsu method, then carrying out binarization, and finally carrying out edge detection and morphological processing to obtain the position of a damaged area.
4. Detecting and identifying wheel tread damage, and realizing automatic identification of tread damage types by adopting an image template matching and feature extraction algorithm;
5. the method comprises the steps of tread image segmentation, wherein image preprocessing operation is simultaneously carried out on a plurality of tread images to obtain segmented tread regions, and then gray level transformation is carried out to enable the average gray level of the plurality of images to be the same; and adjusting on the basis of gray level change, and selecting each segmented tread area as a spliced original image according to the actual splicing requirement.
6. And (3) edge fitting: firstly, finding out a rim edge line of a segmented tread image through an edge detection algorithm, then fitting edges, fitting bent edges into vertical edges, removing the influence of tread radian, and avoiding the concave-convex feeling of a spliced image;
7. and splicing the images of all the sections, namely, finishing image seamless splicing by utilizing corresponding characteristic points or edges in an overlapped public view field region between the images by combining the image acquisition speed aiming at the fitted images, and performing image smoothing treatment at the splicing position of the two images by adopting a gray weighted average algorithm.
Best mode this example is a further description of the second embodiment, and in step four, the least squares fit 1/4 is used to fit the edges of the image of the wheel tread.
Specifically, 1/4 wheel tread's image is at first through edge detection algorithm find segmentation tread image rim edge line, then through least square algorithm fit edge, with crooked edge fit vertical edge, get rid of tread radian influence, avoid the concatenation image to appear unsmooth sense.
Claims (5)
1. The wheel tread damage detection method based on the urban rail vehicle wheel tread damage detection system comprises a wheel sensor (1), a control module (2), an image acquisition control and processing module (3) and a camera module (4);
the wheel sensor (1) is electrically connected with the control module (2) and used for generating a shooting trigger signal when a wheel passes by and sending the shooting trigger signal to the control module (2);
the control module (2) is electrically connected with the image acquisition control and processing module (3) and is used for generating a shooting command according to a shooting trigger signal and sending the shooting command to the image acquisition control and processing module (3);
the image acquisition control and processing module (3) is electrically connected with the camera module (4) and is used for sending a shooting command to the camera module (4);
the camera module (4) is used for acquiring partial images of the wheel tread after receiving a shooting command, and acquiring multiple partial images of the wheel tread for the same wheel; the partial image of the wheel tread surface is sent to an image acquisition control and processing module (3);
the image acquisition control and processing module (3) is also used for splicing the multiple wheel tread surface partial images of the same wheel into a complete wheel tread surface after receiving the wheel tread surface partial images;
the camera module (4) comprises a front wheel shooting camera set (4-1) and a rear wheel shooting camera set (4-2);
the front wheel shooting camera set (4-1) comprises a plurality of pairs of front wheel tread image shooting cameras (4-1-1), and each pair of front wheel tread image shooting cameras (4-1-1) are symmetrically distributed on the outer sides of the two tracks and are respectively used for collecting wheel tread partial images of front wheels on two sides;
the distances between each pair of front wheel tread image shooting cameras (4-1-1) are equal;
the rear wheel shooting camera set (4-2) comprises a plurality of pairs of rear wheel tread image shooting cameras (4-2-1), and each pair of rear wheel tread image shooting cameras (4-2-1) are symmetrically distributed on the outer sides of the two rails and are respectively used for collecting wheel tread partial images of rear wheels on two sides;
and the distances between each pair of rear wheel tread image shooting cameras (4-2-1) are equal;
the wheel sensor (1) comprises a front wheel trigger module group (1-1) and a rear wheel trigger module group (1-2);
the front wheel trigger module group (1-1) comprises front wheel trigger modules (1-1-1) the number of which is the same as that of the front wheel tread image shooting cameras (4-1-1), each front wheel trigger module (1-1-1) corresponds to each pair of front wheel tread image shooting cameras (4-1-1) one by one, and each front wheel trigger module (1-1-1) generates a shooting trigger signal corresponding to the front wheel tread image shooting camera (4-1-1) when a wheel passes by and sends the shooting trigger signal to the control module (2);
the rear wheel trigger module group (1-2) comprises rear wheel trigger modules (1-2-1) the number of which is the same as that of the rear wheel tread image shooting cameras (4-2-1), each rear wheel trigger module (1-2-1) corresponds to each pair of rear wheel tread image shooting cameras (4-2-1) one by one, and each rear wheel trigger module (1-2-1) generates a shooting trigger signal corresponding to the rear wheel tread image shooting camera (4-2-1) when a wheel passes by and sends the shooting trigger signal to the control module (2);
the number of the front wheel trigger modules (1-1-1) is m, and the logarithm of the front wheel tread image shooting camera (4-1-1) is m pairs; the distance between m pairs of front wheel tread image shooting cameras (4-1-1) is 1/m times of the wheel tread circumference;
each front wheel trigger module (1-1-1) is arranged at a position which can shoot 1/m times of wheel tread circumference in front of the corresponding front wheel tread image shooting camera (4-1-1); wherein m is an integer of 3 or more;
the shooting command comprises a front wheel instant shooting command and a front wheel delayed shooting command;
the lens direction of the front wheel tread image shooting cameras (4-1-1) faces the rear of the running of the vehicle, and when each front wheel tread image shooting camera (4-1-1) receives a corresponding front wheel instant shooting command, the collected partial image of the wheel tread is 1/m of the wheel tread; when a corresponding front wheel delayed shooting command is received, the collected wheel tread partial image is 1/2m of the wheel tread;
and the m front wheel tread image shooting cameras (4-1-1) at the same side shoot sequentially along with the advancing of the vehicle, and at least m continuous wheel tread partial images are collected;
the method is characterized by comprising the following specific steps:
step one, image acquisition: each front wheel tread image shooting camera (4-1-1) and each rear wheel tread image shooting camera (4-2-1) shoot to obtain 1/4 wheel treads and 1/8 wheel treads, and 4 front wheel tread image shooting cameras (4-1-1) and 4 rear wheel tread image shooting cameras (4-2-1) on the same side send the shot wheel tread partial images of each wheel to the image acquisition control and processing module (3);
step two, image preprocessing: determining a partial image saliency region of the wheel tread, cutting and correcting the 1/4 wheel tread to obtain a 1/4 wheel tread in a rectangular form, and extracting a 1/4 wheel tread accurate region by using a 1/8 wheel tread by adopting an image projection method;
step three, segmenting partial image of the wheel tread: simultaneously operating multiple 1/4 wheel treads to obtain a segmented tread area, and then performing gray level transformation to ensure that the average gray level of the multiple 1/4 wheel treads are the same; then selecting each segmented tread area according to the actual splicing requirement;
step four, edge fitting: firstly, finding out a rim edge line of a segmented tread area image, then fitting edges, fitting bent edges into vertical edges, and removing the influence of tread radian of a wheel to obtain a fitted image;
step five, splicing each section of image: aiming at the fitted images, according to the image acquisition speed, utilizing corresponding characteristic points or edges in the overlapped public view field areas among the images to finish 1/4 wheel tread seamless splicing to obtain a complete wheel tread;
step six, roughly positioning damage of the wheel tread: obtaining a gray threshold, carrying out binarization on the complete wheel tread, and carrying out edge detection and morphological processing to obtain a damaged area position;
seventhly, detecting and identifying damage of the wheel tread: and realizing automatic identification of the tread damage category by adopting an image template matching and feature extraction algorithm.
2. The method of detecting damage to a wheel tread according to claim 1,
in step four, the least squares fit 1/4 is used to fit the edges of the image of the wheel tread.
3. The method of detecting damage to a wheel tread according to claim 1,
the number of the rear wheel trigger modules (1-2-1) is n, and the logarithm of the rear wheel tread image shooting camera (4-2-1) is n pairs; the distance between the multiple pairs of rear wheel tread image shooting cameras (4-2-1) is 1/n of the wheel tread circumference;
each rear wheel trigger module (1-2-1) is arranged at a position behind the corresponding rear wheel tread image shooting camera (4-2-1) and capable of shooting 1/2n wheel tread circumferences; wherein n is an integer of 3 or more.
4. The method of detecting damage to a wheel tread according to claim 3,
the shooting command also comprises a rear wheel instant shooting command and a rear wheel delayed shooting command;
the orientation of a lens of the rear wheel tread image shooting camera (4-2-1) is the same as the advancing direction of the vehicle, and when each rear wheel tread image shooting camera (4-2-1) receives a corresponding rear wheel delay shooting command, the collected partial image of the wheel tread is 1/n of the wheel tread; when a corresponding rear wheel instant shooting command is received, the collected wheel tread partial image is 1/2n of the wheel tread;
and the n rear wheel tread image shooting cameras (4-2-1) on the same side shoot sequentially along with the advancing of the vehicle and at least collect n continuous wheel tread partial images.
5. The method of detecting damage to a wheel tread according to claim 1,
the wheel sensor (1) also comprises far-end magnetic steel (1-3);
the far-end magnetic steel (1-3) is electrically connected with the control module (2) and is used for sending a vehicle receiving signal to the control module (2) when a vehicle passes by; the vehicle-sending device is also used for sending a vehicle-sending signal to the control module (2) when the vehicle arrives and leaves;
the control module (2) is also electrically connected with the door opening and closing device of the rail side chassis and is used for opening the door opening and closing device of the rail side chassis after receiving the vehicle receiving signal; the door opening and closing device is also used for closing the rail-side chassis after receiving the car sending signal;
the camera module (4) is arranged in the rail-side case.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011040267.5A CN112109769B (en) | 2020-09-28 | 2020-09-28 | Wheel tread damage detection method based on urban rail vehicle wheel tread damage detection system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011040267.5A CN112109769B (en) | 2020-09-28 | 2020-09-28 | Wheel tread damage detection method based on urban rail vehicle wheel tread damage detection system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112109769A CN112109769A (en) | 2020-12-22 |
| CN112109769B true CN112109769B (en) | 2021-07-16 |
Family
ID=73797878
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011040267.5A Active CN112109769B (en) | 2020-09-28 | 2020-09-28 | Wheel tread damage detection method based on urban rail vehicle wheel tread damage detection system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112109769B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114572273B (en) * | 2022-03-15 | 2024-04-30 | 南京拓控信息科技股份有限公司 | Railway vehicle wheel set tread 3D image detection method |
| CN115535027B (en) * | 2022-12-01 | 2023-04-14 | 成都盛锴科技有限公司 | Wheel quality judgment method and system for rail vehicle |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100999219A (en) * | 2006-12-18 | 2007-07-18 | 杭州电子科技大学 | On-line detection method and device for thread defect of vehicle wheel set |
| CN101913367A (en) * | 2009-07-20 | 2010-12-15 | 北京福斯达轨道交通技术有限公司 | Railway wheel tread scratch image dynamic detection device |
| CN105203552A (en) * | 2015-09-18 | 2015-12-30 | 南京拓控信息科技有限公司 | 360-degree tread image detecting system and method |
| CN108520521A (en) * | 2017-04-20 | 2018-09-11 | 南京航空航天大学 | The method of wheel tread extraction and splicing based on image procossing |
| CN110228503A (en) * | 2019-06-21 | 2019-09-13 | 中国神华能源股份有限公司 | Vehicle tyre tread image synthesis system and method |
-
2020
- 2020-09-28 CN CN202011040267.5A patent/CN112109769B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100999219A (en) * | 2006-12-18 | 2007-07-18 | 杭州电子科技大学 | On-line detection method and device for thread defect of vehicle wheel set |
| CN101913367A (en) * | 2009-07-20 | 2010-12-15 | 北京福斯达轨道交通技术有限公司 | Railway wheel tread scratch image dynamic detection device |
| CN105203552A (en) * | 2015-09-18 | 2015-12-30 | 南京拓控信息科技有限公司 | 360-degree tread image detecting system and method |
| CN108520521A (en) * | 2017-04-20 | 2018-09-11 | 南京航空航天大学 | The method of wheel tread extraction and splicing based on image procossing |
| CN110228503A (en) * | 2019-06-21 | 2019-09-13 | 中国神华能源股份有限公司 | Vehicle tyre tread image synthesis system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112109769A (en) | 2020-12-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | A review of applications of visual inspection technology based on image processing in the railway industry | |
| CN109878552B (en) | Foreign matter monitoring devices between track traffic platform door and train based on machine vision | |
| CN112109769B (en) | Wheel tread damage detection method based on urban rail vehicle wheel tread damage detection system | |
| CN104608799B (en) | Based on information fusion technology Railway wheelset tread damage on-line checking and recognition methodss | |
| CN211731391U (en) | Intelligent online detection device for track surface defects | |
| CN110047070B (en) | Method and system for identifying rail wear degree | |
| CN109238756B (en) | Dynamic image detection equipment and detection method for freight car operation fault | |
| CN111198371A (en) | Forward-looking obstacle detection system | |
| CN104331910A (en) | Track obstacle detection system based on machine vision | |
| CN105203552A (en) | 360-degree tread image detecting system and method | |
| CN108520521A (en) | The method of wheel tread extraction and splicing based on image procossing | |
| CN112132896A (en) | Trackside equipment state detection method and system | |
| CN105893953A (en) | Method and system for detecting two license plates of one vehicle | |
| CN108564627B (en) | Linear array image sleeper positioning and counting method based on multi-region gray projection | |
| CN111292294A (en) | Method and system for detecting abnormality of in-warehouse bottom piece | |
| CN110688903B (en) | Barrier extraction method based on train AEB system camera data | |
| CN108830822B (en) | Pantograph carbon skateboard abrasion identification method based on improved Canny operator | |
| CN109934135B (en) | Rail foreign matter detection method based on low-rank matrix decomposition | |
| CN114581437A (en) | Visual inspection system for visual track guided by visual identification | |
| CN112733709A (en) | Track inspection detection system and method | |
| CN110084171B (en) | Detection device and detection method for foreign matters on top of subway train | |
| CN111855667A (en) | Novel intelligent train inspection system and detection method suitable for metro vehicle | |
| CN102937413A (en) | Brake shoe abrasion detection device based on image processing technology | |
| CN113591643A (en) | Underground vehicle station entering and exiting detection system and method based on computer vision | |
| Espino et al. | Rail and turnout detection using gradient information and template matching |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |