A non-contact rail transit train speed direction mileage detection method and system

Abstract

The present invention discloses a non-contact rail transit train speed direction mileage detection method and system, including setting a speed measurement pattern on the train axle in advance, installing an image acquisition unit at the bottom of the train body and focusing on the speed measurement pattern; using the image acquisition unit to collect the rotation image data of the axle speed measurement pattern when the train is running in real time and at high speed, and transmitting it to the image data processing unit through the network for image recognition processing; according to the difference between the recognition processing results of the two frames of images before and after and the acquisition time difference between the two frames of images before and after, the vehicle running direction, the rotation angle and number of turns of the train axle are calculated, and the running mileage and real-time running speed of the train are calculated in combination with the wheel diameter. The present invention has a wide range of applications, does not require the modification of the running parts of the train axle, has high speed measurement accuracy, timely speed update, higher cumulative calculation accuracy of mileage, and can accurately detect the running direction of the vehicle.

Description

Technical Field

[0001] The present invention relates to the field of rail transit technology, and in particular to a machine vision-based contactless rail transit train running speed, running direction and running range measurement method and system. Background Art

[0002] With the rapid development of urban rail transit and state-owned electrified railways in my country, the state of the contact network and track is a key factor affecting the safe operation of electrified trains. Therefore, a series of contact network and track detection devices installed on "rail vehicles and operating trains" have been developed. All kinds of detection devices need to obtain vehicle running speed information in real time to realize the positioning of detection data.

[0003] Generally, the vehicle running speed is obtained by adding speed sensors to the train axles. For safety reasons, operating trains are not allowed to make any changes to the axles. Speed ​​information is generally obtained by analyzing the vehicle TCMS / MVB, but its refresh frequency is slow, generally 1 time / s, and cannot provide accurate speed information for the detection device.

[0004] For example, the patent application with application number CN201810365144.5 discloses a new type of train speed and distance measuring equipment and method, which includes a high-frequency camera, an image recognition module, a logic processing module, and a data interface; wherein the high-frequency camera takes photos of the track vertically downward at fixed intervals, and uses the periodically distributed ground structure between the tracks, the image recognition module and the ratio of the actual and photographed objects to calculate the displacement difference between adjacent pictures; the logic processing module calculates the running speed according to the high-frequency camera shooting frequency and the displacement difference calculated by the image recognition module, performs a logic check according to the performance of the train and the shooting cycle, and the data interface communicates with the on-board equipment to output and input corresponding information. Although this solution can quickly measure the speed of the train relative to the ground; it is easily affected by the external environment during the image acquisition and recognition process, and the measurement result has a large error with the actual data, which needs further improvement.

[0005] For example, the patent application with application number CN201010510191.8 discloses a vehicle speed measurement device based on track characteristics, including a measurement device, which senses changes in track characteristics and collects signals; a measurement device adjustment unit, which adjusts the position of the measurement device to ensure that the measurement device is within the effective measurement range; and a data acquisition and processing unit, which receives the signals collected by the measurement device, and calculates the train operation parameters in real time after signal processing. The device of this scheme is based on a non-contact measurement probe, which obtains parameters such as train running distance, movement speed, acceleration and deceleration through real-time acquisition and processing of the appearance or image features of track sleepers, fasteners, etc., and stores, communicates and displays them. Although this scheme can also realize non-contact measurement of train speed, it is also obtained based on the processing of track characteristics, so it is also easily affected by the external environment, and the measurement results have a large error with the actual data, which needs further improvement. Summary of the invention

[0006] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a contactless rail transit train speed direction mileage detection method and system, which automatically identifies the edge of the train axle pattern by using an image edge detection algorithm, calculates the train axle rotation angle and number of circles by the change in the edge distance of the axle pattern in the two frames of images before and after, and calculates the real-time running speed of the train by combining the wheel diameter and time difference.

[0007] The objective of the present invention is achieved through the following technical solutions:

[0008] A non-contact rail transit train speed direction mileage detection method comprises the following steps:

[0009] Step 1: Set a speed measurement pattern on the train axle in advance, install the image acquisition unit at the bottom of the train body and focus on the speed measurement pattern;

[0010] Step 2: using the image acquisition unit to collect the rotation image data of the axle speed measurement pattern of the train in real time and at high speed, and transmit it to the image data processing unit through the network for image recognition processing;

[0011] Step 3: Based on the difference between the recognition and processing results of the two frames of images and the time difference between the two frames of images, the vehicle direction, the rotation angle and number of turns of the train axle are calculated, and the train mileage and real-time running speed are calculated in combination with the wheel diameter.

[0012] The speed measurement pattern is a linear triangle or trapezoidal pattern with a height equal to the circumference of the train axle, and the number of patterns is composed of one or more identical patterns.

[0013] Specifically, step 2 includes:

[0014] S21: using an image acquisition unit to acquire an image of a speed measurement pattern on an axle when the train is running in real time, to form a high-definition image of the rotating axle;

[0015] S22: filtering the axle high-definition image by using Gaussian filtering to remove image noise, thereby obtaining a high-definition image of the axle after noise reduction processing;

[0016] S23: The finite difference of the first-order partial derivative is used to calculate the amplitude and direction of the speed measurement pattern gradient in the axle high-definition image, and the gradient amplitude is non-maximum suppressed;

[0017] S24: The dual threshold algorithm detects and connects the edges to identify the edges on both sides of the speed measurement pattern in the axle high-definition image after non-maximum suppression, and calculates the distance between the left and right edges on both sides of the speed measurement pattern.

[0018] Specifically, step S24 includes: using a dual threshold algorithm to detect and connect edges to identify and process the axle high-definition image after non-maximum suppression; setting a fixed image height y in the axle high-definition image, selecting edge points on the fixed image height y to obtain a left edge point P1 (x1, y) and a right edge point P2 (x2, y); and calculating the lateral distance X between points P1 and P2.

[0019] Specifically, step three includes:

[0020] S31: The image acquisition unit acquires the speed measurement pattern on the axle at a high speed according to a preset acquisition frequency to obtain a high-definition image stream of the axle, and processes the high-definition image stream of the axle according to the image recognition processing process in step 2 to obtain n lateral distances X1, X2, ..., X n ;

[0021] S32: Obtain the image acquisition times T1, T2, ..., T corresponding to the n lateral distances X by using the timestamp of the axle high-definition image or the system time when the axle high-definition image is transmitted to the image data processing unit. n ;

[0022] S33: Then, the wheel diameter d and the lateral distance X are calculated. max With X min The difference between the two is calculated as X max To X min The running distance L of the current train in the time period, and the total running mileage of the train is calculated by accumulating the running distance L during the running time of the train;

[0023] S34: Based on the lateral distance X corresponding to the two frames of images before and after the current train n , X n-1 , running time T n and T n-1Calculate the current train speed S. The speed S calculation formula is:

[0024]

[0025] S35: At the same time, through this X n Compared with last time n-1 The positive or negative difference of X can be used to obtain the direction of vehicle movement, that is, when X increases, it is forward movement, and when X decreases, it is reverse movement;

[0026] S36: When the speed measurement pattern in the high-definition image of the axle is at the maximum and minimum critical points, it means that the wheel has rotated a full circle, and the accumulated distance error within one circle of the wheel is corrected. At the same time, X n -X n-1 The calculation requires the use of X max , that is, the change value of the axle lateral distance X is X n +X max -X n-1 .

[0027] A contactless rail transit train speed direction mileage detection system is used to implement the above-mentioned contactless rail transit train speed direction mileage detection method, including: an image acquisition unit and an image data processing unit, the image data processing unit is connected to the image acquisition unit; the image acquisition unit is used to high-speed acquire the rotation image data of the train axle speed measurement pattern when the train is running, and upload it to the image data processing unit; the image data processing unit is used to receive, analyze and process the axle rotation image data, and calculate the train running speed, running direction and running mileage in real time.

[0028] Among them, the image acquisition unit includes a high-definition camera and a fill light device; the high-definition camera and the fill light device are installed at the bottom of the train through the C-shaped groove of the train body; and the image data processing unit is connected to the high-definition camera.

[0029] Furthermore, a circle of linear triangle or trapezoidal (one or more) speed measurement patterns with a height equal to the circumference of the train axle is pasted on the train axle.

[0030] Beneficial effects of the present invention:

[0031] 1. The method of the present invention adopts a non-contact method based on machine vision to realize the train running speed detection. It only needs to install the equipment on the existing C-slot interface at the bottom of the train body, and there is no need to modify the train axle running parts, which greatly ensures the safety of rail train operation.

[0032] 2. The measuring method and device of the present invention are widely applicable and can be applied to engineering vehicles such as overhead line inspection vehicles, overhead line operation vehicles, track inspection vehicles, track operation vehicles, as well as urban rail transit operating trains, large railway operating trains, high-speed EMU trains and other models.

[0033] 3. The non-contact speed measurement method of the present invention has a high acquisition frequency, that is, the speed measurement accuracy is high, the speed update is timely, the cumulative calculation accuracy of the mileage is higher, and the running direction of the vehicle can be accurately detected. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flow chart of the method of the present invention.

[0035] Figure 2 It is a system composition architecture diagram of the present invention.

[0036] Figure 3 It is a schematic diagram of the system measurement principle of the present invention.

[0037] Figure 3 Middle: 1-HD camera, 2-speed measurement pattern, 3-train axle. DETAILED DESCRIPTION

[0038] In order to have a clearer understanding of the technical features, purposes and beneficial effects of the present invention, the technical solution of the present invention is now described in detail below. Obviously, the implementation cases described are part of the embodiments of the present invention, not all of the embodiments, and cannot be understood as limiting the scope of the implementation of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of the present invention.

[0039] The existing method generally obtains the vehicle running speed by installing a speed sensor on the vehicle axle; for safety reasons, operating trains are not allowed to make any changes to the axle, and generally obtain speed information by analyzing the vehicle TCMS / MVB, but its refresh frequency is relatively slow, generally 1 time / s, and cannot provide accurate speed information for the detection device. In view of the above shortcomings, the present invention designs a non-contact rail transit train speed direction mileage detection method and system, which automatically identifies the edge of the train axle pattern by using an image edge detection algorithm, calculates the train axle rotation angle and number of circles based on the change in the edge distance of the axle pattern of the two frames of images, and combines the wheel diameter and time difference to calculate the real-time running speed of the train and other information. The detailed design process of the present invention is shown in the following embodiments.

[0040] Embodiment 1:

[0041] In this embodiment, Figure 1 As shown, since the axle of a rail transit train rotates synchronously with the wheels when the train is running, a non-contact train running speed measurement method based on machine vision is designed according to this characteristic, that is, a non-contact rail transit train posture measurement method, which includes the following steps:

[0042] Step 1: Set a speed measurement pattern on the train axle in advance, install the image acquisition unit at the bottom of the train and focus on the speed measurement pattern;

[0043] Step 2: using the image acquisition unit to collect the rotation image data of the axle speed measurement pattern of the train in real time and at high speed, and transmit it to the image data processing unit through the network for image recognition processing;

[0044] Step 3: Based on the difference between the recognition and processing results of the two frames of images and the time difference between the two frames of images, the vehicle direction, the rotation angle and number of turns of the train axle are calculated, and the train mileage and real-time running speed are calculated in combination with the wheel diameter.

[0045] In this embodiment, the pattern setting process of the train axle includes pasting (or painting) a circle of black and white linear triangles or trapezoids (1 or more) patterns on the axle of the rail transit train to facilitate the calculation of the rotation distance of the train wheels.

[0046] In this embodiment, step 2 specifically includes:

[0047] S21: Using an image acquisition unit to acquire images of the speed measurement pattern on the axle when the train is running in real time, to form a high-definition image of the rotating axle.

[0048] S22: Filter the axle high-definition image using Gaussian filtering to remove image noise, thereby obtaining a denoised axle high-definition image.

[0049] Gaussian smoothing filter is a linear filter used to eliminate Gaussian noise and is widely used in image processing for noise reduction. Gaussian filtering is to perform weighted averaging on the entire image. The value of each pixel is obtained by weighted averaging the value of itself and other pixels in the neighborhood.

[0050] S23: The finite difference of the first-order partial derivative is used to calculate the amplitude and direction of the speed measurement pattern gradient in the axle high-definition image, and the gradient amplitude is non-maximum suppressed.

[0051] After the image is smoothed by Gaussian filtering, some of the image's Gaussian noise will be reduced and the sharpness will also be weakened, but at this time the image's edge decomposition is not very obvious, the grayscale change is not very strong, and it is not easy to capture. Therefore, in order to define the size of this edge grayscale change, the image gradient is proposed, and the size of the gradient is used to represent the size of the edge grayscale change.

[0052] The gradient is a vector (vector), which indicates that the directional derivative of a function at that point has the maximum value along that direction, that is, the function changes fastest along that direction (the direction of this gradient) at that point, and the rate of change is the largest (the modulus of the gradient). That is, the derivative (gradient). For an image, the image is regarded as a function f(x,y)f(x,y)f(x,y), and the rate of change of the image is represented by differentials. Since the image is a two-dimensional function, partial derivatives are required. To find the gradient (amplitude and direction) at any point, the Soble convolution operator (optional size, generally 3×3) is used to convolve the corresponding pixel point. For the gradient of edge points, if edge supplementation is not performed, the surrounding pixels are defaulted to 0.

[0053] Since the image edge calculated by the gradient is very fuzzy, there are many edge gradient responses. We strive to find a local maximum value in a gradient direction to represent the gradient response in that gradient direction. The advantage of doing this is that it is easy to calculate and understand. The gradient directions in each area are all approximated as one gradient direction, and the maximum value is sought before and after this approximated gradient direction.

[0054] S24: The dual threshold algorithm detects and connects the edges to identify the edges on both sides of the speed measurement pattern in the axle high-definition image after non-maximum suppression, and calculates the distance between the left and right edges on both sides of the speed measurement pattern.

[0055] Among them, in step S24, the specific implementation process of dual threshold detection and edge connection includes:

[0056] (1) Select appropriate high and low thresholds based on the image. Usually, the high threshold is 2 to 3 times the low threshold.

[0057] (2) If the gradient value of a pixel is higher than the high threshold, it is retained;

[0058] (3) If the gradient value of a pixel is lower than the low threshold, it is discarded;

[0059] (4) If the gradient value of a pixel is between the upper and lower thresholds, the pixel gradient value is searched from the 8-neighborhood of the pixel. If there is a pixel gradient value higher than the upper threshold, it is retained; if not, it is discarded.

[0060] Then, a fixed image height y is set in the axle high-definition image, and edge points on the fixed image height y are selected to obtain the left edge point P1 (x1, y) and the right edge point P2 (x2, y); the lateral distance X between the two points P1 and P2 is calculated.

[0061] In this embodiment, step three specifically includes:

[0062] S31: The image acquisition unit acquires the speed measurement pattern on the axle at a high speed according to a preset acquisition frequency to obtain a high-definition image stream of the axle, and processes the high-definition image stream of the axle according to the image recognition processing process in step 2. Under high-speed acquisition (200 Hz and above), a large number of X, i.e., n lateral distances X1, X2, ..., X n ;

[0063] S32: Obtain the image acquisition times T1, T2, ..., T corresponding to the n lateral distances X by using the timestamp of the axle high-definition image or the system time when the axle high-definition image is transmitted to the image data processing unit. n ;

[0064] S33: Then, the wheel diameter d and the lateral distance X are calculated. max With X min The difference between the two is calculated as X max To X min The running distance L of the current train in the time period, and the total running mileage of the train is calculated by accumulating the running distance L during the running time of the train;

[0065] S34: Based on the lateral distance X corresponding to the two frames of images before and after the current train n , X n-1 , running time T n and T n-1 Calculate the current train speed S. The speed S calculation formula is:

[0066]

[0067] S35: At the same time, through this X n Compared with last time n-1 The positive or negative difference of is used to obtain the direction of vehicle movement, that is, when X increases, it is forward movement (the difference is positive), and when X decreases, it is reverse movement (the difference is negative);

[0068] S36: When the speed measurement pattern in the high-definition image of the axle is at the maximum and minimum critical points, it means that the wheel has rotated a full circle, and the accumulated distance error within one circle of the wheel is corrected. At the same time, X n -X n-1 The calculation requires the use of X max , that is, the change value of the axle lateral distance X is X n +X max -X n-1 .

[0069] Embodiment 2:

[0070] In this embodiment, in order to implement the non-contact rail transit train posture measurement method in the first embodiment, a non-contact rail transit train speed direction mileage detection system is designed, such as Figure 2 As shown, the system includes: an image acquisition unit and an image data processing unit, the image data processing unit is connected to the image acquisition unit; the image acquisition unit is used to collect the rotation image data of the train axle 3 when the train is running at high speed, and upload it to the image data processing unit; the image data processing unit is used to receive, analyze and process the axle rotation image data, and calculate the train running speed, running direction and running mileage in real time.

[0071] The image acquisition unit includes a high-definition camera 1 and a fill-light device; the high-definition camera 1 is installed on the image data processing unit through a C-shaped slot and is connected to the high-definition camera 1 and the fill-light device.

[0072] Furthermore, a circle of linear triangular or trapezoidal (one or more) speed measurement patterns with a height equal to the circumference of the train axle 3 is pasted on the train axle 3.

[0073] In this embodiment, the image data processing unit is an industrial computer, and image processing software can be set on the industrial computer to implement the train posture measurement method of the first embodiment.

[0074] In this embodiment, the image acquisition unit (HD camera 1 and fill light device) installed on the vehicle body focuses on the speed measurement pattern 2 (a linear triangle or trapezoid (1 or more) pattern with distinct black and white) on the train axle 3. The HD camera 1 synchronously triggers the fill light device to collect high-definition images of the train at high speed. The image data processing unit (industrial computer) receives, analyzes and processes the axle rotation image data, and calculates the train running speed in real time. The principle diagram of the measurement system is shown in FIG. Figure 3 shown.

[0075] The detailed measurement principle and method of this embodiment are as follows:

[0076] 1. Train axle pattern: Paste (or paint) a circle of black and white linear triangles or trapezoids (1 or more) on the axle of the rail transit train.

[0077] 2. Image acquisition unit: The image acquisition unit consists of a high-definition camera 1 and a fill light device. The high-definition camera 1 triggers the fill light device to work synchronously to acquire images. The image acquisition unit is installed at the bottom of the train body and focuses on the train axle pattern, and acquires a high-definition image of the pattern on the train axle in real time.

[0078] 3. Image data processing unit (industrial computer): The high-definition images of the train axles collected above are transmitted to the image data processing host (industrial computer) via a gigabit network. The edge detection algorithm is used to identify the edges on both sides of the pattern and calculate the distance between the left and right edges.

[0079] 4. Mileage, direction and speed calculation: Based on the above steps, the train axle rotation angle and number of turns can be calculated according to the difference in the edge distance of the axle pattern identified by the two frames of image recognition and the time difference between the two frames. The running mileage and real-time running speed of the train can be calculated by combining the wheel diameter. The running direction of the vehicle can be obtained by the positive and negative difference in the edge distance of the two frames. Provide accurate mileage positioning information for the detection device.

[0080] This embodiment has the following technical advantages:

[0081] 1. This method uses a non-contact method based on machine vision to detect the train running speed. It only needs to install the equipment on the existing C-slot interface at the bottom of the car body, without the need to modify the train axle running parts, which greatly ensures the safety of rail train operation.

[0082] 2. This measurement method and device have a wide range of applications and can be applied to engineering vehicles such as overhead line inspection vehicles, overhead line operation vehicles, track inspection vehicles, track operation vehicles, as well as urban rail transit operating trains, large railway operating trains, high-speed EMU trains and other models.

[0083] 3. The non-contact speed measurement method implemented in this embodiment has a high acquisition frequency, that is, the speed measurement accuracy is high, the speed update is timely, the cumulative calculation accuracy of the mileage is higher, and the running direction of the vehicle can be accurately detected.

[0084] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of the present invention to be protected is defined by the attached claims and their equivalents.

Claims

1. A non-contact rail transit train speed direction mileage detection method, characterized in that: include: Step 1: Set a speed measurement pattern on the train axle in advance, install the image acquisition unit at the bottom of the train body and focus on the speed measurement pattern; Step 2: using the image acquisition unit to collect the rotation image data of the axle speed measurement pattern of the train in real time and at high speed, and transmit it to the image data processing unit through the network for image recognition processing; Step 3: Calculate the vehicle running direction, the rotation angle and number of turns of the train axle according to the difference between the recognition processing results of the two frames and the acquisition time difference of the two frames, and calculate the running mileage and real-time running speed of the train in combination with the wheel diameter; The step 2 specifically includes: S21: using an image acquisition unit to acquire an image of a speed measurement pattern on an axle when the train is running in real time, to form a high-definition image of the rotating axle; S22: filtering the axle high-definition image by using Gaussian filtering to remove image noise, thereby obtaining a high-definition image of the axle after noise reduction processing; S23: The finite difference of the first-order partial derivative is used to calculate the amplitude and direction of the speed measurement pattern gradient in the axle high-definition image, and the gradient amplitude is non-maximum suppressed; S24: The dual threshold algorithm detects and connects the edges to identify the edges on both sides of the speed measurement pattern in the axle high-definition image after non-maximum suppression, and calculates the distance between the left and right edges on both sides of the speed measurement pattern; The step three specifically includes: S31: The image acquisition unit acquires the speed measurement pattern on the axle at a high speed according to a preset acquisition frequency to obtain a high-definition image stream of the axle, and processes the high-definition image stream of the axle according to the image recognition processing process in step 2 to obtain n lateral distances X1, X2, ..., X n ; S32: Obtain the image acquisition times T1, T2, ..., T corresponding to the n lateral distances X by using the timestamp of the axle high-definition image or the system time when the axle high-definition image is transmitted to the image data processing unit. n ; S33: Then, the wheel diameter d and the lateral distance X are calculated. max With X min The difference between the two is calculated as X max To X min The running distance L of the current train in the time period, and the total running mileage of the train is calculated by accumulating the running distance L during the running time of the train; S34: Based on the lateral distance X corresponding to the two frames of images before and after the current train n , X n-1 , running time T n and T n-1 Calculate the current train speed S. The speed S calculation formula is: S35: At the same time, through this X n Compared with last time n-1 The positive or negative difference of X can be used to obtain the direction of vehicle movement, that is, when X increases, it is forward movement, and when X decreases, it is reverse movement; S36: When the speed measurement pattern in the high-definition image of the axle is at the maximum and minimum critical points, it means that the wheel has rotated a full circle, and the accumulated distance error within one circle of the wheel is corrected. At the same time, X n -X n-1 The calculation requires the use of X max , that is, the change value of the axle lateral distance X is X n +X max -X n-1 ; The speed measurement pattern is a linear triangle with a height equal to the circumference of the train axle; The step S24 specifically includes: using a dual threshold algorithm to detect and connect edges to identify and process the axle high-definition image after non-maximum suppression; setting a fixed image height y in the axle high-definition image, selecting edge points on the fixed image height y to obtain a left edge point P1 (x1, y) and a right edge point P2 (x2, y); and calculating the lateral distance X between points P1 and P2.

2. A non-contact rail transit train speed direction mileage detection system, used to implement the non-contact rail transit train speed direction mileage detection method according to claim 1, characterized in that: include: An image acquisition unit and an image data processing unit, wherein the image data processing unit is connected to the image acquisition unit; The image acquisition unit is used to collect the rotation image data of the train axle speed measurement pattern when the train is running at high speed, and upload it to the image data processing unit; the image data processing unit is used to receive, analyze and process the axle rotation image data, and calculate the train running speed, running direction and running mileage in real time.

3. A non-contact rail transit train speed direction mileage detection system according to claim 2, characterized in that: The image acquisition unit includes a high-definition camera and a fill-light device; the high-definition camera and the fill-light device are installed at the bottom of the train through a C-shaped groove of the train body; and the image data processing unit is connected to the high-definition camera.

4. A non-contact rail transit train speed direction mileage detection system according to claim 2, characterized in that: A circle of linear triangles with a height equal to the circumference of the train axle is pasted on the train axle.

Images