A video stream-based interlocking signal acquisition system and intelligent identification method

Abstract

The application discloses a kind of interlocking signal acquisition system and intelligent identification method based on video stream, acquisition system includes computer interlocking system, train dispatching command system, interlocking signal and driving log information acquisition device, the interlocking signal and driving log information acquisition device include: first VGA distributor, second VGA distributor and high-definition video encoder;Interlocking signal acquisition system is divided out a channel for video information transmission by VGA distributor, replaces by shooting the display of computer interlocking system to obtain video information, greatly reduces or shields internal and external interference.Interlocking signal intelligent identification method contains system initialization, signal change detection and signal area type matching, signal state is identified based on signal area type, the processing and record of interlocking signal and other steps, realize the intelligent, accurate, efficient identification of interlocking signal based on video stream.

Description

Technical Field

[0001] This invention relates to the field of information acquisition and recognition, specifically a video stream-based chain signal acquisition system and intelligent recognition method. Background Technology

[0002] With the continuous development of my country's railway industry, the number of railway lines, railway vehicles, and grassroots dispatching and maintenance personnel under the jurisdiction of various railway bureaus is constantly expanding.

[0003] The interlocking signal system currently used by grassroots railway departments is usually the TYJL-ADX V2.8 computer interlocking system from the China Academy of Railway Sciences. It is a mature automatic railway signal control system that improves transportation efficiency and working conditions with economical and reasonable technical measures while ensuring system safety.

[0004] Because the TYJL-ADX computer interlocking system is networked in a closed manner, its information cannot be made available to external services through conventional interface calls, posing a challenge to railway bureaus in customizing personalized services. Existing solutions often involve taking pictures of the terminal screen with a camera, analyzing the images, and then customizing services. However, this method is highly susceptible to interference from various internal and external factors, including screen quality, camera position, lighting, projection by station staff, and differences in office equipment and environments at different stations. Therefore, the accuracy of signal recognition or log information obtained by taking pictures of the terminal screen is difficult to guarantee. Furthermore, due to the differences between railway bureau stations, solutions designed for one station are difficult to promote in other stations with varying environments. Even in stations with similar environments, subtle differences can lead to significant recognition errors. Summary of the Invention

[0005] The purpose of this invention is to provide a video stream-based chain signal acquisition system and intelligent recognition method. This invention first uses a VGA splitter to create a dedicated channel for video information transmission, then encodes the video using a network video encoder to form a standard video stream, which is then handed over to an intelligent host for intelligent analysis and recognition of chain signals.

[0006] Technical solution:

[0007] This invention first discloses a video stream-based interlocking signal acquisition system, which includes a computer interlocking system, a train dispatching and command system, and an interlocking signal and train log information acquisition device, wherein:

[0008] The computer interlocking system includes: a first server and a first display;

[0009] The train dispatching and command system includes: a second server and a second display;

[0010] The interlocking signal and vehicle log information acquisition device includes: a first VGA splitter, a second VGA splitter, and a high-definition video encoder; the first VGA splitter and the second VGA splitter are both one-input two-output video splitters, each with one VGA input terminal, one VGA output terminal, and one HDMI output terminal; the high-definition video encoder is configured with two inputs and one output, having two high-definition input terminals, HDMI1 and HDMI2, and one ETH Ethernet output port.

[0011] The input of the first VGA distributor is connected to the VGA output of the first server of the computer interlocking system. One output of the first VGA distributor is connected to the first monitor, and the other output is connected to the HDMI1 high-definition input of the high-definition video encoder. The input of the second VGA distributor is connected to the VGA output of the second server of the train dispatching and command system. One output of the second VGA distributor is connected to the second monitor, and the other output is connected to the HDMI2 high-definition input of the high-definition video encoder. The ETH Ethernet output of the high-definition video encoder is connected to the intelligent host or server. The intelligent host or server receives the data from the high-definition video encoder and performs interlocking signal and train log information identification applications.

[0012] The high-definition video encoder encodes high-definition video from its HDMI1 and HDMI2 inputs based on H.625, forming RTMP / RTSP stream data which is then output through the ETH Ethernet output port.

[0013] The intelligent host receives RTMP / RTSP stream data from the high-definition video encoder and performs chain signal and vehicle log information recognition applications.

[0014] Preferably, the computer interlocking system selected is the TYJL-ADX V2.8 computer interlocking system;

[0015] Preferably, VGA splitter 1 and VGA splitter 2 are selected from CAIBEIKE VGA to VGA+HDMI converters;

[0016] Preferably, the high-definition video encoder is the XE2D from Chengdu Zhongtian Zhiguang Technology Co., Ltd.

[0017] The present invention also discloses a method for intelligent recognition of chain signals based on video streams, the steps of which include: system initialization, signal change detection and signal region type matching, signal status recognition based on signal region type, and processing and recording of chain signals.

[0018] System initialization refers to the system reading the configuration file and initializing each region object that needs to be identified according to the configuration file. The region object records the signal status of each region at the time of initialization and the time of signal recording using keywords.

[0019] During system initialization, the system reads a configuration file that combines each signal identification region with the actual signal name of the site, assigning practical meaning to the calibrated regions. This prepares for the later automatic generation of site-specific information based on the identified signals. The calibration of each signal identification region can be performed manually according to the actual site or using a trained image segmentation model. The configuration file uses JSON format, with the following structure:

[0020]

[0021] Here, `shapes` represents the complete set of information for all regions. Each shape within the set contains information about a specific region, including Label, Name, and Points; ... indicates that a `shapes` set includes multiple shapes, the specific number of which is configured according to the actual site conditions.

[0022] Label is the tag for this location information. Depending on the type of signal identification area, the tag naming format is L_nnn, S_nnn, X_nnn. L_ indicates that the area is a railway line, S_ indicates a signal, B_ indicates a signal button, X_ indicates a turnout area; nnn is an Arabic numeral number.

[0023] Name is the actual signal name at the field station, which is used for automatic signal generation later;

[0024] Points is a set of points consisting of the two-dimensional coordinates of the outline points of the target area, which are composed of the upper left and lower right coordinates of each type of area;

[0025] For the entire site, there is a Flash Rate configuration item with a flash threshold. The setting value m represents the selected frequency, which is used to determine whether the signal status is flashing later.

[0026] In signal change detection and signal region type matching, the positional grayscale change value obtained by calculating the frame difference frame by frame is used to detect whether the signal interface of the current interlocking system has changed. If a change is found, the location of the signal change is determined based on the grayscale change position. The system finds the signal region information corresponding to that location according to the configuration file, calls the corresponding recognition algorithm for the region type, and performs signal state recognition for the entire region. Signal regions that have not changed are not processed, improving signal recognition efficiency.

[0027] The specific process for identifying signal status based on signal region type is as follows:

[0028] If the signal area type is a railway line, button, or turnout area, the pixel color distribution of the entire signal area is statistically analyzed. The signal status of the current area is then determined based on the color distribution of the signal area. For button areas, two special cases are identified: gray buttons and red buttons. Gray buttons represent "Train dispatching in progress," and red buttons represent "Corresponding area blocked."

[0029] If the signal area type is a signal controller, then two steps are required: color distribution analysis and graphic judgment of the signal area.

[0030] During color distribution analysis, the current area is divided into single-signal light areas and dual-signal light areas based on the aspect ratio of the signal region. For a single-signal light area, the RGB values ​​of 10 pixels in its center area are extracted and averaged to determine the signal light's acceptance. For a dual-signal light area, the center positions of the two signal lights are first determined according to their aspect ratio, and then the RGB values ​​of 10 pixels from each signal light are extracted and averaged to determine the colors of the two signal lights.

[0031] The image recognition uses a straight line detection algorithm to extract straight line information in the current area. Based on the slope of the lines, horizontal lines are filtered out. The start and end points of the line segments contained within the line are used to determine whether the line belongs to the border line, and then the color of the border line is extracted. The signal status of the traffic light is determined by comprehensively considering the traffic light color, the presence of border lines, and the border color. The specific traffic light statuses identified include: red light, yellow light, green light, blue light, green-yellow light, double yellow light, blue light with white border, and red light with green border.

[0032] The specific process for processing and recording chain signals is as follows:

[0033] For areas where signal changes occur, the signal status will be updated, the time difference between the last signal recording time and the current time will be calculated, and the current signal recording time will be updated.

[0034] The difference between the current signal update time and the last update time of the area object is compared with the flicker threshold to determine whether the area is flickering. If it is determined to be flickering, the current area status is updated to flickering status again.

[0035] Based on the current status of the signal area and the actual business logic on site, the system will automatically generate railway dispatching and maintenance terminology that is consistent with the actual global signal meaning and present it to the on-site staff for review, modification and release.

[0036] The beneficial effects of this invention are:

[0037] 1) The present invention provides a video stream-based interlocking signal acquisition system and intelligent recognition method, which uses a VGA splitter to split a dedicated channel for video information transmission, replacing the acquisition of video information by shooting the monitor of the computer interlocking system, greatly reducing or shielding the interference of external factors such as screen image quality, camera position, lighting, projection of site staff, and differences in office equipment and environment of each site on the video signal.

[0038] 2) The video is encoded using a network video encoder to form a standard video stream, which is then sent to an intelligent host or server for intelligent identification of interlocking signals. This effectively improves the accuracy of signal identification.

[0039] 3) The system detects whether the signal interface of the interlocking system has changed by calculating the frame difference value frame by frame. The location of the signal change is determined based on the position of the grayscale change. The system finds the signal region information corresponding to that location according to the configuration file, calls the corresponding recognition algorithm for the region type, and performs signal status recognition for the entire region. Signal regions that have not changed are not processed, effectively reducing the amount of recognition computation, lowering power consumption, and improving signal recognition efficiency.

[0040] 4) The configuration file settings lay the foundation for signal type matching and provide a target for different stations to identify based on video streams, greatly improving the system deployment efficiency.

[0041] 5) The video stream-based intelligent chain signal recognition method classifies signal area types into railway lines, buttons, switches, and stop signals, and performs area signal type matching before recognizing the specific type, which greatly improves the signal recognition accuracy and work efficiency. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of a video stream-based interlocking signal acquisition system according to the present invention.

[0043] Figure 2 This is a flowchart of a video stream-based intelligent recognition method for chain signals according to the present invention.

[0044] Figure 3 This is an example of a railway line and its identification results in the interlocking system of the present invention.

[0045] Figure 4 This is an example of a turnout in the interlocking system of the present invention, along with a diagram showing the identification results.

[0046] Figure 5 This is an example of a button and a diagram showing the recognition results in the chain system of the present invention.

[0047] Figure 6 This is an example of a signal controller and its recognition result diagram in the interlocking system of the present invention. Detailed Implementation

[0048] The present invention will be further described below with reference to specific embodiments.

[0049] like Figure 1 As shown, a video stream-based interlocking signal acquisition system includes a computer interlocking system, a train dispatching and command system, and interlocking signal and train log information acquisition devices, wherein:

[0050] The computer interlocking system includes: a first server and a first display;

[0051] The train dispatching and command system includes: a second server and a second display;

[0052] The interlocking signal and vehicle log information acquisition device includes: a first VGA splitter, a second VGA splitter, and a high-definition video encoder; the first VGA splitter and the second VGA splitter are both one-input two-output video splitters, each with one VGA input terminal, one VGA output terminal, and one HDMI output terminal; the high-definition video encoder is configured with two inputs and one output, having two high-definition input terminals, HDMI1 and HDMI2, and one ETH Ethernet output port.

[0053] The input of the first VGA distributor is connected to the VGA output of the first server of the computer interlocking system. One output of the first VGA distributor is connected to the first monitor, and the other output is connected to the HDMI1 high-definition input of the high-definition video encoder. The input of the second VGA distributor is connected to the VGA output of the second server of the train dispatching and command system. One output of the second VGA distributor is connected to the second monitor, and the other output is connected to the HDMI2 high-definition input of the high-definition video encoder. The ETH Ethernet output of the high-definition video encoder is connected to the intelligent host or server. The intelligent host or server receives the data from the high-definition video encoder and performs interlocking signal and train log information identification applications.

[0054] The high-definition video encoder encodes high-definition video from its HDMI1 and HDMI2 inputs based on H.625, forming RTMP / RTSP stream data which is then output through the ETH Ethernet output port.

[0055] The intelligent host receives RTMP / RTSP stream data from the high-definition video encoder and performs chain signal recognition applications.

[0056] Specifically, in this application, the computer interlocking system is selected as the TYJL-ADX V2.8 computer interlocking system; the VGA distributor 1 is selected as the CAIBEIKE VGA to VGA+HDMI converter; and the high-definition video encoder is selected as the XE2D from Chengdu Zhongtian Zhiguang Technology.

[0057] like Figure 2 As shown, a method for intelligent recognition of chain signals based on video streams includes the following steps: system initialization, signal change detection and signal region type matching, signal state recognition based on signal region type, and processing and recording of chain signals.

[0058] System initialization refers to the system reading the configuration file and initializing each region object that needs to be identified according to the configuration file. The region object records the signal status of each region at the time of initialization and the time of signal recording using keywords.

[0059] During system initialization, the system reads a configuration file that combines each signal identification region with the actual signal name of the site, assigning practical meaning to the calibrated regions. This prepares for the later automatic generation of site-specific information based on the identified signals. The calibration of each signal identification region can be performed manually according to the actual site or using a trained image segmentation model. The configuration file uses JSON format, with the following structure:

[0060]

[0061] Here, `shapes` represents the complete set of information for all regions. Each shape within the set contains information about a specific region, including Label, Name, and Points; ... indicates that a `shapes` set includes multiple shapes, the specific number of which is configured according to the actual site conditions.

[0062] Label is the tag for this location information. Depending on the type of signal identification area, the tag naming format is L_nnn, S_nnn, X_nnn. L_ indicates that the area is a railway line, S_ indicates a signal, B_ indicates a signal button, X_ indicates a turnout area; nnn is an Arabic numeral number.

[0063] Name is the actual signal name at the field station, which is used for automatic signal generation later;

[0064] Points is a set of points consisting of the two-dimensional coordinates of the outline points of the target area, which are composed of the upper left and lower right coordinates of each type of area;

[0065] For the entire site, there is a Flash Rate configuration item with a flash threshold. The setting value m represents the selected frequency, which is used to determine whether the signal status is flashing later.

[0066] For example:

[0067]

[0068]

[0069]

[0070] Flash Rate: 2# indicates that the flashing threshold frequency of the signal lights in this area is 2 times / second.

[0071] In signal change detection and signal region type matching, the positional grayscale change value obtained by calculating the frame difference frame by frame is used to detect whether the signal interface of the current interlocking system has changed. If a change is found, the location of the signal change is determined based on the grayscale change position. The system finds the signal region information corresponding to that location according to the configuration file, calls the corresponding recognition algorithm for the region type, and performs signal state recognition for the entire region. Signal regions that have not changed are not processed, improving signal recognition efficiency.

[0072] The specific process for identifying signal status based on signal region type is as follows:

[0073] If the signal area type is a railway line, button, or turnout area, the pixel color distribution of the entire signal area is statistically analyzed. The signal status of the current area is then determined based on the color distribution of the signal area. For button areas, two special cases are identified: gray buttons and red buttons. Gray buttons represent "no signal shunting terminal," and red buttons represent "the area corresponding to the button is blocked."

[0074] Examples and identification results of railway lines in the interlocking system are as follows: Figure 3 As shown, the turnout example and recognition results are as follows: Figure 4 As shown, button examples and recognition results are as follows. Figure 5 As shown.

[0075] If the signal area type is a signal controller, then two steps are required: color distribution analysis and graphic judgment of the signal area.

[0076] During color distribution analysis, the current area is divided into single-signal light areas and dual-signal light areas based on the aspect ratio of the signal region. For a single-signal light area, the RGB values ​​of 10 pixels in its center area are extracted and averaged to determine the signal light's acceptance. For a dual-signal light area, the center positions of the two signal lights are first determined according to their aspect ratio, and then the RGB values ​​of 10 pixels from each signal light are extracted and averaged to determine the colors of the two signal lights.

[0077] The graphic judgment uses a line detection algorithm to extract line information in the current area. Based on the slope of the lines, horizontal lines are filtered out. The start and end points of the line segments contained within the line are used to determine whether the line belongs to the border line, and then the color of the border line is extracted. The signal status of the traffic light is determined by combining the color of the traffic light, the presence of border lines, and the border color.

[0078] Example of a signal controller and recognition results in a chain system: Figure 6 As shown

[0079] The specific process for processing and recording chain signals is as follows:

[0080] For areas where signal changes occur, the signal status will be updated, the time difference between the last signal recording time and the current time will be calculated, and the current signal recording time will be updated.

[0081] The difference between the current signal update time and the last update time of the area object is compared with the flicker threshold to determine whether the area is flickering. If it is determined to be flickering, the current area status is updated to flickering status again.

[0082] Based on the current status of the signal area and the actual business logic on site, the system will automatically generate railway dispatching and maintenance terminology consistent with the actual global signal meaning, as required by the customer, and present it to on-site staff for review, modification, and publication. Specifically, based on the overall identification status of railway lines, signals, buttons, and switches in different signal areas, and based on the currently implemented entry, exit, passage, and shunting business logic on site, the system will automatically generate railway dispatching and maintenance terminology consistent with the actual global signal meaning, as required by the customer. (For example, when executing the departure of a train from a certain station, if the exit signal is identified as "green-yellow," the system will generate "There are two block sections ahead of the station; the train can exit.") This will be presented to on-site staff for review, modification, and publication.

[0083] As Figure 3 The first example will generate "Railway line 7G idle"; if... Figure 4 The second example will generate "Switch 13 is in the reverse position"; for example... Figure 5 The fourth example will generate a "button-corresponding area blocked" message; for example... Figure 6 The seventh legend will generate "Signal blue and white box shunting route start".

[0084] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. A method for intelligent recognition of chain signals based on video streams, characterized in that, The method is based on a video stream-based interlocking signal acquisition system, including a computer interlocking system, a train dispatching and command system, and an interlocking signal and train operation log information acquisition device, wherein: The computer interlocking system includes: a first server and a first display; The train dispatching and command system includes: a second server and a second display; The interlocking signal and vehicle log information acquisition device includes: a first VGA splitter, a second VGA splitter, and a high-definition video encoder; the first VGA splitter and the second VGA splitter are both one-input two-output video splitters, each with one VGA input terminal, one VGA output terminal, and one HDMI output terminal; the high-definition video encoder is configured with two inputs and one output, having two high-definition input terminals, HDMI1 and HDMI2, and one ETH Ethernet output port. The input of the first VGA distributor is connected to the VGA output of the first server of the computer interlocking system. One output of the first VGA distributor is connected to the first monitor, and the other output is connected to the HDMI1 high-definition input of the high-definition video encoder. The input of the second VGA distributor is connected to the VGA output of the second server of the train dispatching and command system. One output of the second VGA distributor is connected to the second monitor, and the other output is connected to the HDMI2 high-definition input of the high-definition video encoder. The ETH Ethernet output of the high-definition video encoder is connected to the intelligent host or server. The intelligent host or server receives the data from the high-definition video encoder and performs interlocking signal and train log information identification applications. The method steps include: system initialization, signal change detection and signal region type matching, signal state identification based on signal region type, and processing and recording of chain signals; the specific process of identifying signal state based on signal region type is as follows: If the signal area type is a railway line, button, or turnout area, then the pixel color distribution of the entire signal area is statistically analyzed; the signal status of the current area is determined based on the color distribution of the signal area; for button areas, two special cases are identified: gray buttons and red buttons. Gray buttons represent "vehicle dispatching in progress", and red buttons represent "corresponding area blocked". If the signal area type is a signal controller, then two steps are required to analyze the signal area: color distribution and graphic judgment. When performing color distribution analysis, the current area is divided into single-signal light areas and dual-signal light areas according to the aspect ratio of the signal area. For single-signal light areas, the RGB values ​​of 10 pixels in the center area are extracted and averaged to determine the signal light acceptance. For dual-signal light areas, the center positions of the two signal lights are first determined according to the aspect ratio, and then the RGB values ​​of 10 pixels in each area are extracted and averaged to determine the colors of the two signal lights. The image judgment uses a line detection algorithm to extract line information in the current area. Based on the slope of the line, horizontal lines are filtered out. Based on the start and end points of the line segments contained in the line, it is determined whether the line belongs to the border line, and then the color of the border line is extracted. The signal status of the signal controller is determined by a combination of factors, including the color of the signal light, the presence of a straight border line, and the color of the border line. In the signal change detection and signal region type matching, the position grayscale change value obtained by calculating the frame difference frame by frame is used to detect whether the signal interface of the current interlocking system has changed; if it has changed, the position of the signal change is determined according to the position of the grayscale change; the system finds the signal region information corresponding to the position according to the configuration file, calls the corresponding recognition algorithm for the region type, and performs signal status recognition for the entire region; no processing is performed on the signal regions corresponding to the unchanged regions, thereby improving the signal recognition efficiency.

2. The method according to claim 1, characterized in that, The system initialization refers to the system reading the configuration file and initializing each region object that needs to be identified according to the configuration file. The region object records the signal status of each region at the time of initialization and the time of signal recording using keywords.

3. The method according to claim 2, characterized in that, The system initialization process involves the system reading a configuration file that combines each signal identification area with the actual signal name of the station, giving the calibration area a practical meaning. This prepares for the automatic generation of station-customized information based on the identified signals in the later stages. The calibration of each signal identification area is performed manually according to the actual station or by using a trained image segmentation model.

4. The method according to claim 1, characterized in that, The specific process for processing and recording the chain signal is as follows: For areas where signal changes occur, the signal status will be updated, the time difference between the last signal recording time and the current time will be calculated, and the current signal recording time will be updated. The difference between the current signal update time and the last update time of the area object is compared with the flicker threshold to determine whether the area is flickering. If it is determined to be flickering, the current area status is updated to flickering status again. Based on the current status of the signal area and the actual business logic on site, the system will automatically generate railway dispatching and maintenance terminology that is consistent with the actual global signal meaning and present it to the on-site staff for review, modification and release.

Images