A train carriage segmentation method based on line scan imaging
Through the magnetic steel matching line scanning camera, image acquisition is triggered by using magnetic steel signals, which solves the image quality and algorithm complexity problems of the existing train segmentation method, and achieves an efficient and real-time cabin segmentation accuracy of 1.2mm.
Patent Information
- Application Number
- CN202111331052.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-11-11
- Publication Date
- 2025-08-26
- Estimated Expiration
- 2041-11-11
AI Technical Summary
The existing train segmentation methods have problems such as high image quality requirements, complex segmentation algorithms and high error segmentation rate, especially for high-speed rail carriages that cannot be implemented.
The magnetic steel is used to combine the linear scanning camera imaging method, and the line numbers obtained by the linear scanning camera are specially marked and stored, and the image acquisition is triggered by the magnetic steel signal, and the car segmentation position is determined to achieve real-time and high-precision segmentation.
It realizes efficient car segmentation without the processing of the upper computer algorithm, with a segmentation accuracy of 1.2mm, which is suitable for real-time segmentation during train operation.
Smart Images

Figure CN116128884B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of rail transit automated measurement, and in particular to a method for segmenting train carriages using line-scan imaging. Background Art
[0002] In the field of rail transit, using visual inspection technology to analyze train surface features is an important means of intelligent train inspection. Since trains travel at high speeds and are long and narrow, high-resolution imaging is required. Therefore, line scan imaging is an effective way to scan the body of the vehicle and obtain images. To determine the specific location of the detection features on the body, the carriages in the image need to be segmented.
[0003] The current methods for train segmentation are: 1. Identify the line scan image based on the features (bellows or couplers) at the car segmentation point, and use the feature position as the segmentation position; 2. Use laser and photoelectric sensors to shoot at both sides of the track to sense the gaps between the car connections and perform car segmentation; Among them, method 1 has high requirements on image quality, segmentation algorithm, and feature recognition algorithm, and the detection time is long, otherwise it is easy to cause problems such as mis-segmentation; method 2 cannot be implemented for trains with seamless bellows at the car connection, such as high-speed rail cars. Summary of the Invention
[0004] In order to solve the above technical problems, this method adopts the method of using magnetic steel in combination with line scan camera for imaging. By specially marking and storing the row numbers obtained by the line scan camera, the image row number corresponding to when the wheel triggers the magnetic steel is accurately known, and then the number of the image row where the segmentation position is located is obtained. This method does not require the host computer to perform additional image algorithm processing, has high segmentation efficiency, and can obtain the car segmentation position image row in real time during the train operation. The segmentation accuracy can reach 1.2mm.
[0005] The technical solution is as follows:
[0006] A method for segmenting carriages using train line scan imaging is provided. At least two magnets are provided along the train track. The magnet farthest from the oncoming train direction is designated as magnet A, and the magnet second farthest from the oncoming train direction is designated as the trigger magnet. The signal emitted when a train wheel pair passes magnet A is designated as a marking signal, and the signal emitted when a train wheel pair passes the trigger magnet is designated as a trigger signal.
[0007] A line scan camera is fixed around the magnet A to scan and image the area to be inspected; the track length between the installation position of the line scan camera and the trigger magnet is greater than the distance between the first set of wheels and the front of the train;
[0008] The line scan camera and each magnet are connected to the controller;
[0009] When the first set of wheels of the train passes the trigger magnet, the controller receives a trigger signal and sends a start signal to the line scan camera, which then captures images at a specified line frequency. Simultaneously, the controller stores the images and labels each line of the image in time sequence.
[0010] When any wheelset passes the magnetic steel A, the controller receives a marking signal, marks the image row L captured at this time, records its number, and stores the corresponding number with the sequence number of the current wheelset in the train;
[0011] Based on the known train model, find the serial numbers of a pair of adjacent wheelsets that are not installed under the same car, and then determine the labels of the corresponding rows L, marked as i and j respectively. Use the following formula to obtain the label k of the image row corresponding to the car segmentation position on the image;
[0012]
[0013] In the formula, when the magnet A is closer to the direction of the oncoming vehicle than the line scan camera, the sign is positive, and when the magnet A is farther away from the direction of the oncoming vehicle than the line scan camera, the sign is negative.
[0014] The same method is used to obtain the label k of the image row corresponding to all the compartment segmentation positions on the image to complete the compartment segmentation.
[0015] Furthermore, there are one or more line scan cameras.
[0016] Preferably, the length of the track between the line scan camera and the magnetic steel A is less than 30 cm.
[0017] Preferably, the line connecting the line scanning camera and the center of the magnetic steel A is not perpendicular to the direction of travel of the train.
[0018] Furthermore, the line scan camera captures images at a specified line frequency, and the specified line frequency is obtained by:
[0019] When the same set of wheels passes by at least two magnets, the train's running speed is obtained based on the time difference between the magnets sending signals to the controller and the distance difference between the two magnets;
[0020]
[0021] Preferably, 3 to 6 groups of magnetic steels are provided, and the other magnetic steels are used for real-time speed measurement of the train.
[0022] Furthermore, the method for determining the serial numbers of adjacent wheelsets that are not installed under the same carriage is as follows:
[0023] The train model is obtained by using the vehicle number image collected on the train body or setting up a radio frequency reader next to the track. According to the train model, the number of wheelsets corresponding to each car is known, and then the serial numbers of adjacent wheelsets that are not installed under the same car are known (the serial numbers of the two wheelsets corresponding to the front and rear of the car connection).
[0024] Furthermore, the line scan camera is a two-dimensional line array scan camera or a three-dimensional line scan camera.
[0025] This method uses the marking signal emitted by the wheelset passing through the magnetic steel A to record the image row L, obtain the label of the image row L, determine the image row L corresponding to the front and rear wheelset at the car connection, and then accurately calculate the image of the segmentation position. The multiple sets of magnetic steel can also be used for train speed measurement. The acquisition line frequency of the line scan camera is adjusted according to the real-time travel speed of the train to improve the accuracy of the image segmentation position. The present invention does not require the host computer to perform additional image algorithm processing, has high segmentation efficiency, and can obtain the car segmentation position image row in real time during the train operation. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 Schematic diagram of the system structure in a specific implementation manner. DETAILED DESCRIPTION
[0027] The technical solution of the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.
[0028] A train carriage segmentation method based on line scan imaging, such as Figure 1 As shown, at least two magnets are provided along the train track. The magnet farthest from the oncoming direction is denoted as magnet A, and the magnet second farthest from the oncoming direction is denoted as trigger magnet. The signal emitted when the train wheels pass magnet A is denoted as the marking signal, and the signal emitted when the train wheels pass the trigger magnet is denoted as the trigger signal.
[0029] A line scan camera is fixed around the magnet A to scan and image the area to be inspected. To ensure that the line scan camera can capture the image of the train head in a timely manner, the track length between the line scan camera installation position and the trigger magnet is greater than the distance between the first set of wheels of the train and the train head.
[0030] The line scan camera and each magnet are connected to the controller;
[0031] When the first set of wheels of the train passes the trigger magnet, the controller receives a trigger signal and sends a start signal to the line scan camera, which then captures images at a specified line frequency. The line scan camera captures one line of image per frame, where each line of image is a column of pixels in the vertical direction (perpendicular to the direction of travel) of the train.
[0032] At the same time, the controller stores the image and labels each row of the image in time sequence;
[0033] When any wheelset passes the magnetic steel A, the controller receives a marking signal, marks the image row L captured at this time, records its number, and stores the corresponding number with the sequence number of the current wheelset in the train;
[0034] Based on the pre-known train model, find the serial numbers of a pair of adjacent wheelsets that are not installed under the same carriage (such as Figure 1 The wheelset M and wheelset N in the image are the order of the wheelsets in the train, and then the labels of the corresponding rows L are determined, marked as i and j respectively; the label k of the image row corresponding to the car segmentation position on the image is obtained using the following formula;
[0035] k=((i+j) / 2)±(v / r)×(S / v)
[0036] S represents the length of track between the line scan camera and magnet A, v represents the speed of the first set of wheels of the train when it passes through this track (this speed can be calculated by triggering the magnet and magnet A, or by any other two magnets), and r represents the longitudinal resolution of a single pixel of the line scan camera, where the longitudinal direction is perpendicular to the direction of train travel.
[0037] Right now:
[0038]
[0039] The vertical resolution of a single pixel of the line scan camera is the resolution perpendicular to the train's direction of travel. When magnet A is closer to the oncoming train than the line scan camera, the vertical resolution is positive, and when magnet A is farther away from the oncoming train than the line scan camera, the vertical resolution is negative.
[0040] The same method is used to obtain the label k of the image row corresponding to all the compartment segmentation positions on the image to complete the compartment segmentation.
[0041] Among them, the method for determining the serial numbers of adjacent wheelsets that are not installed under the same carriage is:
[0042] The train model is obtained by using the vehicle number image collected on the train body or setting up a radio frequency reader next to the track. According to the train model, the number of wheelsets corresponding to each car is known, and then the serial numbers of adjacent wheelsets that are not installed under the same car are known (the serial numbers of the two wheelsets corresponding to the front and rear of the car connection).
[0043] The line scan camera can be a two-dimensional line array scan camera or a three-dimensional line scan camera.
[0044] In specific implementation, one or more line scan cameras are fixed around the magnet A according to the train model, the number of areas to be inspected, and their height. More specifically, when there are multiple inspection areas, such as the roof and both sides of the carriage, line scan cameras are set at the corresponding positions to scan and image each area to be inspected.
[0045] The length of the track between any line scan camera and the trigger magnet is greater than the distance between the first set of wheels and the front of the train;
[0046] Each line scan camera is connected to a controller.
[0047] In order to obtain accurate image segmentation, the distance between the position where the line scan camera takes pictures and the position directly above the wheelset should not be too long. In specific implementation, the corresponding track length between the line scan camera and the magnetic steel A is preferably less than 30 cm.
[0048] A more preferred setting is that the line connecting the line scan camera and the center of the magnet A is perpendicular to the direction of the train's travel. For example, if the line scan camera is used to capture the upper surface of the train, its imaging plane is vertically downward and passes through the center of the magnet A. In this case, the length of the track between the line scan camera and the magnet A is zero. At the same time, this setting can also simplify the calculation of the index k.
[0049] In this embodiment, the line scan camera captures images at a specified line frequency, which is obtained by:
[0050] When the same set of wheels passes by at least two magnets, the train's running speed is obtained based on the time difference between the magnets sending signals to the controller and the distance difference between the two magnets;
[0051]
[0052] As a preferred embodiment, 3 to 6 groups of magnets are provided. When there are only 3 groups, one magnet A or a trigger magnet is required for speed measurement. More preferably, there are 4 to 6 groups. The magnets other than the magnet A and the trigger magnet are used for real-time speed measurement of the train.
[0053] The present invention does not require a host computer to perform additional image algorithm processing for image segmentation, and has high segmentation efficiency. It uses magnetic steel triggering + line scanning imaging to perform carriage segmentation, and has higher segmentation accuracy.
[0054] The foregoing descriptions of specific exemplary embodiments of the present invention have been presented for purposes of illustration and description. The foregoing descriptions are not intended to be exhaustive or to limit the invention to the precise forms disclosed, and it is apparent that many variations and modifications are possible in light of the foregoing teachings. The exemplary embodiments have been chosen and described in order to explain the specific principles of the invention and their practical application, thereby enabling others skilled in the art to make and utilize the various exemplary embodiments of the invention and various alternatives and modifications thereof. The scope of the invention is intended to be defined by the appended claims and their equivalents.
Claims
1. A method for segmenting train carriages using line scan imaging, characterized by: At least two magnets are installed along the train track. The magnet farthest from the oncoming train direction is marked as magnet A, and the magnet second farthest from the oncoming train direction is marked as trigger magnet. The signal emitted when the train wheel passes magnet A is marked as the marking signal, and the signal emitted when the train wheel passes the trigger magnet is marked as the trigger signal. A line scan camera is fixed around the magnet A to scan and image the area to be inspected; the track length between the installation position of the line scan camera and the trigger magnet is greater than the distance between the first set of wheels and the front of the train; The line scan camera and each magnet are connected to the controller; When the first set of wheels of the train passes the trigger magnet, the controller receives a trigger signal and sends a start signal to the line scan camera, and the line scan camera collects images at a specified line frequency; At the same time, the controller stores the image and labels each row of the image in time sequence; When any wheelset passes the magnetic steel A, the controller receives a marking signal, marks the image row L captured at this time, records its number, and stores the corresponding number with the sequence number of the current wheelset in the train; Based on the pre-known train model, find the serial numbers of a pair of adjacent wheelsets that are not installed under the same carriage, and then determine the numbers of the corresponding rows L, marked as i and j respectively; Use the following formula to obtain the label k of the image row corresponding to the compartment segmentation position on the image; In the formula, when the magnet A is closer to the direction of the oncoming vehicle than the line scan camera, the sign is positive, and when the magnet A is farther away from the direction of the oncoming vehicle than the line scan camera, the sign is negative. The same method is used to obtain the label k of the image row corresponding to all the compartment segmentation positions on the image to complete the compartment segmentation.
2. The method for segmenting a train carriage using line scan imaging according to claim 1, wherein: There are one or more line scan cameras.
3. The method for segmenting a train compartment using line scan imaging according to claim 1 or 2, wherein: The length of the track between the line scan camera and the magnetic steel A is less than 30 cm.
4. The method for segmenting train carriages using line scan imaging according to claim 1, wherein: The line connecting the line scanning camera and the center of the magnet A is not perpendicular to the direction of train travel.
5. The method for segmenting train carriages using line scan imaging according to claim 1, wherein: The line scan camera acquires images at a specified line rate, which is obtained by:
6. The method for segmenting train carriages using line scan imaging according to claim 1, wherein: There are 3 to 6 groups of magnets, and the other magnets are used to measure the speed of the train in real time.
7. The train carriage segmentation method for line scan imaging according to claim 1, characterized in that: The method for determining the serial numbers of adjacent wheelsets that are not installed under the same carriage is as follows: The train model is obtained by using the vehicle number image collected on the train body or setting up a radio frequency reader next to the track. According to the train model, the number of wheelsets corresponding to each car is known, and the serial numbers of adjacent wheelsets that are not installed under the same car are then known.
8. The method for segmenting train carriages using line scan imaging according to claim 1, wherein: The line scan camera is a two-dimensional line scan camera or a three-dimensional line scan camera.
Citation Information
Patent Citations
Vehicle segmentation method based on railway wagon loading video surveillance system
CN110406566A
Vehicle number and box number identification association system monitoring equipment
CN212302511U