A method for detecting a display state of a station signal

Abstract

The present application relates to a kind of station signal display state detection method, it is related to rail transit field, can be non-contact in the absence of electrical connection or communication, obtain the display state information of station signal.The main method flow is as follows: camera shoots the station computer interlocking display screen station yard picture, corrects the perspective of each frame image and extracts station yard picture image from it;As initialization step, manually obtain the coordinates and number of signal machine in station yard picture image, the coordinates of color reference, determine the threshold space of different display states according to the standard color of different display states of signal machine;Detection process is, to each frame of real-time video stream, image preprocessing, intercept the image of signal machine to be detected, according to the threshold space of each display state, calculate and analyze the display state of each signal machine.The present application can obtain the display state of signal machine without the communication interface provided by station computer interlocking system, does not affect the normal use of station computer interlocking system, improves the security of information acquisition.

Description

Technical Field

[0001] This invention relates to the field of rail transit technology, and in particular to a method for detecting the display status of station signal controllers. Background Technology

[0002] The computer interlocking system in rail transit stations, operated by station staff or the ATS (Automatic Train Control) system, enables mutual constraint and control between turnouts, signals, and track circuits within the station. It is the core equipment for safe operation of rail transit and has the highest level of functional safety.

[0003] Station computer interlocking systems are functional safety systems with strict limitations on connections to other systems, often not providing external communication interfaces. Rail transit departments such as train operations, track maintenance, and rolling stock require access to the display status information of various station signals for equipment maintenance and track work. This allows for a quick understanding of the real-time status of trains and the station yard, facilitating work and ensuring operational safety. To obtain signal display status, the station computer interlocking system typically needs to provide additional communication interfaces, along with the development of corresponding dedicated communication protocols. Due to considerations of potential safety risks and equipment management needs, station computer interlocking systems often require various safety analyses, assessments, and tests before providing external communication interfaces; sometimes, even these interfaces are not available. Therefore, safely obtaining station signal display status is one of the challenges faced by rail transit departments such as train operations and track maintenance. Summary of the Invention

[0004] The purpose of this invention is to overcome the above-mentioned problems and provide a method for detecting the display status of station signal controllers, so as to achieve the goal of quickly and safely detecting the display status of station signal controllers.

[0005] The objective of this invention can be achieved through the following detection method: a method for detecting the display status of station signals, characterized in that a camera is used to capture images of the interlocking display screen at the station site, and the captured images are processed and analyzed to obtain signal information, thereby achieving non-electrical acquisition of the display status of station signals.

[0006] A method for detecting the display status of a station signal controller includes the following steps:

[0007] S1. A camera installed in a fixed position captures images of the station computer interlocking display screen. The captured video footage includes images of the station area. Each frame of the image is extracted from the video stream.

[0008] S2. Use perspective transformation to correct the viewing angle of each frame and extract the station field image.

[0009] S3, position information extraction is performed on the station picture image intercepted in S2, that is, the coordinates of all signal machines in the station picture image are manually obtained and the signal machines are numbered, the coordinates of any M color references are obtained, and the above information is stored. The color references are parts with always unchanged display color in the station picture image;

[0010] S4, according to the R / G / B values r z , g z , b z corresponding to the standard color of different display states of the signal machine and the pre-set error tolerance r allow , g allow , b allow , r z ± r allow , g z ± g allow , b z ± b allow are taken as the threshold space of the display state [( r L, r H ), (g L , g H ), (b L , b H )], and the threshold information is stored. The display states of the signal machine include green, red, yellow and white;

[0011] S5, according to the information stored in S3, image preprocessing is performed on the station picture image, and the signal machine image to be detected is intercepted from the station picture image. The image preprocessing includes filtering and color correction;

[0012] S6, according to the threshold space of different display states, the display state corresponding to the signal machine image to be detected is calculated and analyzed, and the display state of the signal machine is output.

[0013] Further, the coordinates in S3 correspond to a coordinate system with the upper left corner of the station picture image as the origin, the horizontal right as the positive direction of the X axis, the vertical downward as the positive direction of the Y axis, and the pixel point number as the numerical value.

[0014] Further, the specific steps of obtaining the coordinates of all signal machines in the station picture image and numbering the signal machines in S3 are as follows:

[0015] S31, a Hough circle transformation algorithm is used to find all circles and their center coordinates in the station picture image, as the signal machines to be selected, and the center coordinates are identified as the coordinates of the signal machines;

[0016] S32, in the signal machines to be selected, a target signal machine and its coordinates are manually selected and numbered, and the number is recorded as C i, i = 1, 2, …, n, n is the number of signal machines in the station picture image.

[0017] Further, the step S3 is an initialization step, and only one frame of station picture image is used as a sample image, and the information is obtained manually.

[0018] Further, the specific steps of the step S5 are as follows:

[0019] S51, image preprocessing is performed on the station picture image, and the specific steps are as follows:

[0020] S511, R / G / B component images R o , G o , and B o of the station picture image are extracted.

[0021] S512, median filtering is used on each component image to filter out image noise.

[0022] S513, R / G / B values r o , g o , and b o of the color displayed at the corresponding coordinate point in the station picture image are obtained, and each component image is corrected, and the specific formula is as follows:

[0023]

[0024] wherein R a , G a , and B a are the corrected component images, R o , G o , and B o are the uncorrected component images, r s , g s , and b s are R / G / B values of the standard color of the color reference, r o , g o , and b o are R / G / B values of the color displayed by the color reference in the station picture image, and M is the number of color references.

[0025] S514, the component images are combined into a color image.

[0026] S52, the coordinates (x i , y i )(i = 1, 2, …, n) and the number of all signal machines in the station picture image in the step S3 are used to intercept the signal machine image P i to be detected, and the interception range of the signal machine image to be detected is a circular range with the coordinates of the signal machine as the center and a preset radius, and the number C iOne-to-one correspondence storage;

[0027] Further, the step S6 specifically comprises the following steps:

[0028] S61, representing the to-be-detected signal machine image P with a set of R / G / B values i The specific calculation formula is:

[0029]

[0030] Wherein, r i , g i , and b i are the R / G / B values representing the to-be-detected signal machine image P i , N is the total number of pixel points of the to-be-detected signal machine image, r ij , g ij , and b ij are the R / G / B values of the jth pixel point of the to-be-detected signal machine image P i .

[0031] S62, calculating the distance between the to-be-detected signal machine image P i and the threshold space of the display state obtained in the step S4, and the specific calculation formula is:

[0032]

[0033] Wherein, D i is the distance between the to-be-detected signal machine image P i and the threshold space, D ir , D ig , and D ib are the distances between the to-be-detected signal machine image P i and each component of the threshold space, r i , g i , and b i are the R / G / B values representing the to-be-detected signal machine image P i , and (r L , r H , g L , g H , b L , and b H ) are the threshold space of a certain display state.

[0034] The distances between the to-be-detected signal machine image P i and the threshold spaces of the possible various display states of the signal machine are calculated respectively, and a set [D i1 , D i2 , D i3 , D i4, k = 1, 2…, 4, respectively represent four display states, namely green, red, yellow, white.

[0035] S63, find the minimum value D i1 ,D i2 ,D i3 ,D i4 ] in the set [D iz , that is, the image P i of the signal to be detected is the closest to the zth display state Z, and the signal machine display state information in the form of [signal machine number C i + display state Z] is output;

[0036] Compared with the prior art, the present application has the following advantages:

[0037] First, the present application uses a camera to obtain a station field picture image of a computer interlocking display screen of a station, processes and judges the real-time collected station field picture image, so as to obtain a corresponding detection result, thereby realizing the purpose of real-time detection of the display state of a station signal machine in a non-intrusive manner without contact, effectively reducing the detection engineering quantity, improving the detection speed, and ensuring the safety of detection.

[0038] Second, in the present application, the initialization information is artificially calibrated in advance, the image to be detected can be accurately and quickly intercepted from the collected image, the display color of the image to be detected is described by a set of R / G / B values, the state of each signal machine is judged by using the threshold space obtained in advance, and the accuracy of the detection result can be ensured. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 is a method flowchart of the present application;

[0040] Figure 2a is an original sample image in the embodiment;

[0041] Figure 2b is a calibration image in the embodiment;

[0042] Figure 2c is a station field picture image intercepted in the embodiment; DETAILED DESCRIPTION

[0043] The present application will be described in detail below in combination with the drawings and specific embodiments.

[0044] Embodiment

[0045] To realize a station signal machine display state detection method, as shown in Figure 1 , the following steps are included:

[0046] S1, a camera installed at a fixed position captures a station computer interlocking display screen, and a video picture containing a station yard picture is obtained from a video stream, and each frame of image is extracted from the video stream;

[0047] S2, a perspective transformation is used to correct the perspective of each frame of image, and a station yard picture image is extracted;

[0048] S3, position information of the station yard picture image extracted in S2 is extracted, that is, coordinates of all signals in the station yard picture image are manually obtained, and the signals are numbered, and coordinates of any M color references are obtained, and the above information is stored. The color reference is a part of the station yard picture image that always displays a color;

[0049] S4, according to R / G / B values r z 、g z 、b z corresponding to standard colors of different display states of a signal and a pre-set error tolerance value r allow 、g allow 、b allow , r z ±r allow 、g z ±g allow 、b z ±b allow as a threshold space [(r L ,r H ),(g L ,g H ),(b L ,b H )] of the display state, and the threshold information is stored. The display state of the signal includes green, red, yellow and white;

[0050] S5, according to the information stored in S3, image preprocessing is performed on the station yard picture image, and a signal image to be detected is extracted from the station yard picture image. The image preprocessing includes filtering and color correction;

[0051] S6, according to the threshold space of different display states, a display state corresponding to the signal image to be detected is calculated and analyzed, and the display state of the signal is output.

[0052] The embodiment applies the above technical solution, and the specific process includes:

[0053] B100: a camera installed at a fixed position captures a station computer interlocking display screen, and a video picture containing a station yard picture is obtained from a video stream, and each frame of image is extracted from the video stream, as a sample image.

[0054] B200: taking the upper left corner of the sample image as the origin, a coordinate system is established with the horizontal right as the positive direction of the X axis and the vertical downward as the positive direction of the Y axis;

[0055] B300: Artificially correct the perspective of the sample image by perspective transformation, and intercept the station picture image, record the perspective transformation parameters and related coordinates involved for subsequent steps;

[0056] B3001: Calibrate the original sample image F into a calibration image F' with a flat perspective, so that the station picture image in the calibration image is rectangular, the original sample image F is as shown in Figure 2a , the calibration image F' is as shown in Figure 2b , and the specific steps are as follows:

[0057] B30011: Transform each pixel point coordinate in the original sample image F into the corresponding coordinate in the calibration image F' by the transformation calculation formula, and the transformation calculation formula is as follows:

[0058]

[0059] In the formula, (x, y) is the coordinate of the pixel point in the original image, and (x', y') is the corresponding coordinate after transformation;

[0060] B30012: Transform the transformation calculation formula, and obtain the coordinates [(x1, y1), (x2, y2), (x3, y3), (x4, y4)] of the four vertices of the station picture image in the original sample image F by artificial calibration, and set the corresponding coordinates of the four vertices in the calibration image F' as [(x'1, y'1), (x'2, y'1), (x'1, y'2), (x'2, y'2)], and substitute the coordinates of the four vertices before and after transformation into the following matrix:

[0061]

[0062] By solving the above matrix, the parameters a i (i=1, 2, …, 7, 8) required for calibration of the station picture image are obtained.

[0063] In this embodiment, the coordinates of the four vertices in the original sample image F are [(450, 126), (5501, 658), (372, 1244), (5556, 1446)], and the corresponding coordinates of the four vertices in the calibration image F' are [(450, 120), (5500, 120), (450, 1300), (5500, 1300)].

[0064] The calibration parameters a i obtained by this calculation are as follows:

[0065] (6.5187e-1, 5.7700e-2, 1.3851e+2, -1.1826e-1, 1.0570e0, 3.7134e+1, -6.0181e-5, 2.2964e-5)

[0066] The parameter a required for calibration is stored in the data processing unit for use in subsequent steps. i Stored in the data processing unit for use in subsequent steps.

[0067] B3002: Extract the station picture image region R in the calibration image F', the four vertex coordinates [(x'1, y'1), (x'2, y'1), (x'1, y'2), (x'2, y'2)] of the station picture image region R are selected manually in the calibration image F', wherein x'1 < x'2, y'1 < y'2, and the region R is defined as:

[0068] R∈(x,y), x'1≤x≤x'2, y'1≤y≤y'2

[0069] The region R is cut out from the image F' as shown in the figure, and the pixel point with coordinates (x'1, y'1) is set as the coordinate origin. Figure 2c

[0070] In this embodiment, the four vertex coordinates are [(450, 120), (5500, 120), (450, 1300), (5500, 1300)]. The four vertex coordinates are stored in the data processing unit for use in subsequent steps.

[0071] B400: Extract the position information of the cut-out station picture image, manually obtain the coordinates of all signal machines in the station picture image, number the signal machines, obtain the coordinates of any M color references, and store the above information. The color reference is a part of the station picture image such as the station picture background or the platform, which always displays a constant color, and the specific steps are as follows:

[0072] B4001: Use the Hough circle transformation algorithm to find all circles and their center coordinates in the station picture image, as candidate signal machines, and identify the center coordinates as the coordinates of the candidate signal machines;

[0073] B4002: Among the candidate signal machines, manually select the target signal machine and its coordinates, and number them, denoted as C i , i = 1, 2, …, n, n is the number of signal machines in the station picture image;

[0074] B4003: Manually select M points representing the platform or the station picture background, which are characterized by constant content displayed at the location, and store the coordinates of these points.

[0075] ​B500: R / G / B values r z , g z , b z corresponding to the standard color of different display states of the signal machine allow , g allow , b allow and the pre-set error tolerance value r z ± r allow , g z ± g allow , b z ± b allow as the threshold space (r L , r H , g L , g H , b L , b H ) of the display state, store the threshold information. The signal machine display states include green, red, yellow, white;

[0076] B600: According to the information stored in B300, correct the perspective of each frame image of the real-time video stream and intercept the station field image in the same way as the sample image processing procedure;

[0077] B700: Use the coordinates of the M color references stored in B400 to pre-process the station field image, and the specific steps are as follows:

[0078] B7001: Extract the R / G / B component images R o , G o , B o of the station field image

[0079] B7002: Use median filtering on each component image to filter out image noise;

[0080] B7003: Use the color reference coordinates in B400 to obtain the R / G / B values r o , g o , b o of the display color corresponding to the coordinate point in the station field image, and correct each component image, and the specific formula is:

[0081]

[0082] where R a , G a , B a are the corrected component images, B o , G o , B o are the uncorrected component images, r s , g s , bs R / G / B values of the standard color as the color reference, r o , g o , b o R / G / B values of the color displayed in the station picture image by the color reference, M is the number of the color reference.

[0083] B7004: Merge the component images into a color image.

[0084] B800: Use the coordinates (x i ,y i )(i = 1, 2, …, n) and numbers of all the signals in the station picture image described in B400 to intercept the signal image P i to be detected, the interception range of the signal image to be detected is a circle with the coordinates of the signal as the center and a preset value as the radius, and is stored in one-to-one correspondence with the number C i .

[0085] B900: Determine the display state corresponding to the signal image to be detected according to the threshold information of different display states, and output the signal display state:

[0086] B9001: A set of R / G / B values represents the signal image P i to be detected, and the specific calculation formula is:

[0087]

[0088] Wherein, r i , g i , b i are the R / G / B values of the signal image P i to be detected, N is the total number of pixel points of the signal image to be detected, r ij , g ij , b ij are the R / G / B values of the jth pixel point of the signal image P i to be detected.

[0089] B9002: Calculate the distance between the signal image P i to be detected and the threshold space of the display state obtained in step B500, and the specific calculation formula is:

[0090]

[0091] Wherein, D i is the distance between the signal image P i to be detected and the threshold space, D ir , D ig , D ib is the distance between the signal image Pi The distance r from each component of the threshold space i g i b i To represent the image P of the signal machine to be tested i The R / G / B value, (r L ,r H ,g L ,g H ,b L ,b H ) represents the threshold space for a certain display state.

[0092] The image P of the signal machine to be detected was calculated separately. i The distances to the threshold space of all possible display states of the signal are used to obtain the set [D]. i1 D i2 D i3 D i4 k = 1, 2, ..., 4, representing four display states: green, red, yellow, and white.

[0093] B9003: In set [D] i1 D i2 D i3 D i4 Find the minimum value D in ] iz That is, the image P of the signal machine to be detected i The distance to the z-th display state Z is the smallest, and it is defined by [signal number C]. i The signal display status information is output in the form of "+display status Z";

[0094] B1000: Repeat steps B600 to B900 for each frame of the real-time video stream to output the display status of each signal in each frame, thereby realizing the real-time detection function.

[0095] In practical applications, as described in B1000, the camera repeats the above steps for each real-time image frame captured to output the detection result of the display status of each signal in each frame, looping to achieve real-time detection. If a change in the display status of a signal is detected, the camera also records and saves the time of the change, the signal number, the display status before the change, and the display status after the change. In particular, when it is impossible to determine the display status of signal C in the current frame... i The displayed status will still be recorded.

[0096] In summary, this technical solution does not rely on electrical connection with the station's computer interlocking system, and can detect the real-time status of station signals in a contactless manner without direct electrical connection.

[0097] The above specific embodiments do not constitute a limitation on the protection scope of the present application, and the present application can be implemented in other specific forms (such as when the detection object is a display screen station picture of ATS, CTC or other train operation control related systems) without departing from the spirit and scope of the present application. In addition, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application, and these modifications and variations should also be considered as falling within the protection scope of the present application. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

Claims

1. A method for detecting the display status of a station signal, characterized in that, Includes the following steps: (1) A camera installed in a fixed position captures the station computer interlocking display screen. The captured video footage includes the station area. Each frame of the image is extracted from the video stream. (2) Use perspective transformation to correct the viewing angle of each frame and extract the station field image; (3) A coordinate system is established with the upper left corner of the station view image as the origin, the horizontal rightward direction as the positive X-axis, the vertical downward direction as the positive Y-axis, and the number of pixels as the numerical value. This step is an initialization step, which only requires using any frame of the station view image as a sample image and manually obtaining information once. The position information of the station view image captured in (2) is extracted, that is, the coordinates of all signals in the station view image are manually obtained and the signals are numbered. The coordinates of any M color reference objects are obtained and the above information is stored. The color reference objects are the parts of the station view image whose displayed color remains unchanged. (4) The R / G / B value r corresponding to the standard color of the signal for different display states. z g z b z and the preset error tolerance value r allow g allow b allow , with r z ±r allow g z ±g allow b z ±b allow The threshold space for this display state [(r L ,r H ),(g L ,g H ),(b L ,b H The threshold space is stored, and the signal display status includes green, red, yellow, and white. (5) Based on the information stored in step (3), perform image preprocessing on the station screen image and extract the image of the signal to be detected from it. The image preprocessing includes filtering and color correction. (6) Based on the threshold space of different display states, calculate and analyze the display state corresponding to the image of the signal machine to be detected, and output the display state of the signal machine, as follows: Step (6a): Use a set of R / G / B values ​​to represent the signal generator image P to be detected. i The specific calculation formula is as follows: Where, r i g i b i To represent the image P of the signal machine to be tested i The R / G / B values, where N is the total number of pixels in the image of the signal generator to be detected, and r ij g ij b ij Image P of the signal machine to be detected i The R / G / B value of the j-th pixel; Step (6b): Calculate the image P of the signal generator to be detected. i The distance D between the threshold space of the display state obtained in step (4) and the threshold space of the display state. i The specific calculation formula is as follows: In the above formula, D ir D ig D ib Image P of the signal machine to be detected i The calculation formula for the distance to each component of the threshold space is as follows: Where r i g i b i To represent the image P of the signal machine to be tested i The R / G / B value, (r L ,r H ,g L ,g H ,b L ,b H () represents the threshold space for a certain display state; The image P of the signal machine to be detected was calculated separately. i The distances to the threshold space of all possible display states of the signal are used to obtain the set [D]. i1 D i2 D i3 D i4 ], k = 1, 2, ..., 4, representing 4 display states: green, red, yellow, and white; Step (6c): In set [D] i1 D i2 D i3 D i4 Find the minimum value D in ] iz That is, the image P of the signal machine to be detected i The distance to the z-th display state Z is the smallest, and it is defined by [signal number C]. i The signal display status information is output in the form of "+display status Z".

2. The method for detecting the display status of a station signal as described in claim 1, characterized in that, Step (3) involves obtaining the coordinates of all signals in the station's visual image and numbering the signals. This specifically includes the following steps: (2a) Use the Hough circle transform algorithm to find all circles and their center coordinates in the station field image, and use them as candidate signals. The center coordinates are then used as the coordinates of the signal. (2b) From the candidate signals, the target signal and its coordinates are manually selected and numbered, and the number is denoted as C. i , i = 1, 2, ..., n, where n is the number of signals in the station's visual image.

3. The method for detecting the display status of a station signal as described in claim 1, characterized in that, Step (5) specifically includes the following steps: (3a) Image preprocessing of the station view images, the specific steps are as follows: (3a1) Extract the R / G / B component images of the station site images. o G o B o ; (3a2) Apply median filtering to each component image to remove image noise; (3a3) Using the color reference coordinates described in step (3), obtain the R / G / B value r of the color displayed at the corresponding coordinate point in the station image. o g o b o The specific formula for correcting each component image is as follows: Among them, R a G a B a For the corrected component images, R o G o B o To correct the image components before processing, r s g s b s The R / G / B values ​​of the standard color of the color reference object, r o g o b o The R / G / B values ​​of the colors of the color reference objects displayed in the station image are given, where M is the number of color reference objects. (3a4) Merge the component images into a color image; (3b) Use the coordinates (x, y) of all signals in the station view image described in step (3). i ,y i Given the given information (i = 1, 2, ..., n) and their corresponding numbers, extract the image P of the signal generator to be tested. i The cropping range of the image of the signal to be detected is a circle with the coordinates of the signal as the center and the radius as a preset value, and is related to the number C. i One-to-one correspondence storage.

Images