Layered feature fusion-based intelligent analysis method for turnout state of intelligent track-occupying track

Abstract

The invention discloses a layered feature fusion-based intelligent analysis method for an intelligent track switch state of an intelligent track occupation board. The method comprises the following steps of: image preprocessing: carrying out denoising processing on track and switch region images in a video stream; coordinate offset compensation: carrying out coordinate correction on the preprocessed image; color feature extraction: generating a corresponding color mask; shape feature extraction: extracting shape features of the track and the turnout through a contour detection technology; hierarchical feature fusion: fusing the color features and the shape features to generate a comprehensive judgment basis of the track and turnout states; for the state of the turnout, combining the positioning track and the anti-position track to assist in judging the positioning, anti-position and switch-splitting states of the turnout; state output: outputting the final states of the track and the turnout. According to the method, the problems of poor dynamic adaptability and data islands in traditional turnout state judgment are solved through a layered progressive logic architecture and a multi-source data fusion technology.

Description

Technical Field

[0001] The present invention relates to the technical field of rail transportation, and in particular to an intelligent analysis method for track switch states of an intelligent bus stop based on hierarchical feature fusion. Background Art

[0002] The Depot Control Center (DCC) of urban rail transit is responsible for centralized command and dispatch of production operations, ensuring the smooth completion of vehicle inspections, vehicle operation, and comprehensive repair and maintenance of process equipment and facilities. Therefore, the DCC must comprehensively monitor the real-time status of elements such as signals, track lines, switches, and vehicle positions within the depot's microcomputer interlocking system interface, as well as power supply and construction conditions. This allows for timely analysis and early warning of abnormal conditions and illegal operations to prevent production accidents.

[0003] In railway transportation systems, turnouts are critical equipment for guiding trains, and accurate determination of their status is directly related to driving safety and operational efficiency. Traditional turnout status detection methods rely primarily on track circuits, relay activation, and manual inspections. These methods suffer from the following significant issues: Traditional methods fail to fully integrate multi-dimensional information, such as track occupancy, image analysis results, and equipment locking status, resulting in insufficient reliability and real-time performance. Existing technologies also lack effective calibration logic when the track is occupied or in abnormal conditions (such as jammed or blocked), making it difficult to dynamically correct the situation by combining track data and single-lock markings, which can easily lead to safety risks. Summary of the Invention

[0004] The purpose of this section is to summarize some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of this application to avoid obscuring the purpose of this section, the abstract and the title of the invention, and such simplifications or omissions should not be used to limit the scope of the present invention.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a method for intelligent analysis of track switch status using an intelligent busy board based on hierarchical feature fusion, comprising the following steps: S1. Image preprocessing: De-noise the track and switch area images in the video stream, use erosion and dilation operations to remove noise in the signal light area, and optimize image quality through threshold adjustment in the HSV color space; S2. Coordinate offset compensation: According to the offset jitter characteristics of the microcomputer interlocking screen, coordinate correction is performed on the pre-processed image to adapt to the screen offset; S3. Color feature extraction: Generate corresponding color masks based on the color ranges defined by the track and switch states; calculate the number of non-zero pixels in each color mask and determine the color states of the track and switch based on the pixel ratio; S4. Shape feature extraction: The shape features of the track and turnout are extracted using contour detection technology; the single-lock circular feature of the turnout is detected using the convex hull fitting method; S5. Layered feature fusion: Color features are combined with shape features to generate a comprehensive basis for judging track and switch status. For switch status, the occupancy status of the positioning track and the reverse track is combined to assist in judging the positioning, reverse and squeeze status of the switch. S6. Status output: Output the final status of the track and switch based on the fused feature data.

[0006] As a preferred solution of the intelligent analysis method of track switch status of intelligent busy board based on hierarchical feature fusion of the present invention, the color mask generation in S3 includes the following steps: a. Create multiple sets of masks based on preset HSV color ranges (e.g. red, white, blue, pink, yellow); b. Calculate the number of non-zero pixels in each mask group, compare the pixel proportions of each color mask, and select the color state corresponding to the largest proportion as the result of the preliminary judgment;

[0007] As a preferred solution of the intelligent analysis method of track and turnout status of intelligent busy board based on hierarchical feature fusion of the present invention, the judgment of track and turnout in S5 is as follows: a. For track status: a1. If the color feature is blue, the track is considered to be idle; a2. If the color feature is red, it is determined that the track is occupied; a3. If the color feature is white and there is a direction arrow, it is determined that the route is locked; if there is no direction arrow, it is determined to be an abnormal state and a warning is issued; a4. If the color feature shows pink flashing, it is determined that the track is blocked; b. For turnout status: b1. If the turnout name and turnout tip position are displayed in green, it is considered to be positioned; b2. If the turnout name and turnout tip position are displayed in yellow, it is considered to be in reverse position; b3. If there is a green circle at the fork tip, it is considered a single lock; b4. If the turnout name is surrounded by a red box, it is considered blocked; b5. If the switch name is steady red and the switch tip flashes red, it is considered a squeeze switch.

[0008] As a preferred solution of the intelligent analysis method of track switch status based on hierarchical feature fusion of the intelligent busy board of the present invention, the switch status judgment in S5 further includes the following steps: a. When both the positioning and anti-position tracks are idle: a1. Prioritize detecting the red pixel ratio through the red HSV mask. If the red pixel ratio exceeds the threshold, it is judged as a blocked state. If the turnout is not blocked, the shape features of the green and yellow areas are extracted using contour detection technology to further determine the turnout's position and reverse status. b. When the positioning and anti-position orbital states are inconsistent: b1. Determine the initial state of the turnout based on the occupation status of the positioning and counter-position tracks; b2. Use contour detection technology combined with convex hull fitting to detect whether there is a circular feature of a single lock; b3. If there is no single lock feature, the final state of the turnout is determined by comprehensive analysis of color and shape features.

[0009] As a preferred solution of the intelligent analysis method of track switch status based on hierarchical feature fusion of the intelligent bus stop board of the present invention, the image preprocessing in S1 further includes the following steps: a. Adjust the HSV color space for high-noise images and optimize the color threshold range through debugging; b. Perform local filtering on the noise points in the traffic light area to improve the accuracy of the color mask.

[0010] As a preferred solution of the intelligent analysis method of the intelligent busy board track switch state based on hierarchical feature fusion of the present invention, the coordinate offset compensation in S2 includes the following steps: a. Use image registration technology to correct the offset and jitter of the computer interlocking screen in real time; b. Use the fixed feature points of the track or switch as a reference to calculate the coordinate offset and compensate for it.

[0011] As a preferred solution of the intelligent analysis method of the intelligent busy board track switch state based on hierarchical feature fusion of the present invention, the contour detection technology in S4 includes the following steps: a. Extract the outlines of tracks and switches using edge detection algorithms; b. Perform convex hull fitting on the contour to identify the geometric properties of the single-lock circular feature.

[0012] As a preferred solution of the intelligent analysis method of track switch status based on hierarchical feature fusion of the intelligent busbar of the present invention, the status output in S6 includes the following steps: a. Display the final status of the track and turnout in real time on the intelligent busy board in any form of text, color or graphics; b. Transmit status information to the railway control system through the communication interface for safe management of tracks and switches.

[0013] In a second aspect, the present invention further provides an intelligent analysis system for track switch status based on hierarchical feature fusion, comprising: Image acquisition module, used to obtain video streams of track or turnout areas; Image preprocessing module for performing image denoising, coordinate offset compensation, and color mask generation; Feature extraction module, used to extract color features and shape features; Hierarchical feature fusion module, used to combine color features and shape features for state judgment; The status output module is used to output the final status information of the track or switch.

[0014] As a preferred solution of the intelligent analysis system of track switch status based on hierarchical feature fusion of the present invention, the hierarchical feature fusion module further includes: Track status judgment submodule, used to judge the track's idle, occupied, route locked or blocked status based on color characteristics; The turnout status judgment submodule is used to judge the turnout's positioning, reverse position, single lock, blockage or squeezed state by combining color features, shape features and track occupancy status.

[0015] Beneficial effects of the present invention: A multi-source data fusion model is constructed by integrating color mask, contour detection, track occupancy status and other multi-source data. The data weight is dynamically adjusted according to the scenario (for example, track data is prioritized when the track is occupied, and image analysis is prioritized when the track is idle), avoiding the limitations of a single data source and ensuring the reliability of the judgment results.

[0016] In conventional scenarios (idle track), color masks and shape features are used preferentially to quickly determine the switch status (positioning / reversing), with fast response speed and intuitive judgment. In complex scenarios (track occupation, abnormal signals), the calibration logic is automatically triggered, and the judgment results are dynamically corrected based on track occupancy data, single lock marks and historical status to avoid misjudgment caused by failure of a single method. The judgment method is switched according to track status and environmental changes to ensure reliability under complex working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them: Figure 1 This is a flow chart of the intelligent analysis method of track switch status based on intelligent busy board based on hierarchical feature fusion proposed in the present invention; Figure 2 This is a flow chart of the intelligent analysis method of track switch status based on intelligent busy board based on hierarchical feature fusion proposed by the present invention. DETAILED DESCRIPTION

[0018] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0019] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0020] Reference Figure 1-2 The present invention provides a method for intelligent analysis of track switch status using intelligent busy board with hierarchical feature fusion, comprising the following steps: S1. Image preprocessing: De-noise the track and switch area images in the video stream. Use erosion and dilation operations to remove noise in the signal light area. For example, use a 3×3 structuring element and perform two erosion and dilation iterations to preserve the continuity of the target area. Optimize image quality by adjusting the threshold in the HSV color space. The image preprocessing also includes the following steps: a. Adjust the HSV color space for high-noise images and optimize the color threshold range through debugging. Dynamically adjust the HSV color space threshold range for different lighting conditions (such as daytime and nighttime). For example: in daytime scenes, the Hue value range of the red mask is expanded to 0-15 to adapt to high-brightness environments; in nighttime scenes, the Value range of the blue mask is adjusted to 80-200 to enhance blue recognition in low-brightness environments. b. Perform local filtering on the noise points in the traffic light area to improve the accuracy of the color mask; S2. Coordinate offset compensation: Based on the offset jitter characteristics of the microcomputer interlocking screen, coordinate correction is performed on the pre-processed image to adaptively offset the screen. Fixed feature points of the track and switch (such as the fixed axis of the switch, track end points, etc.) are used as reference points. Image registration technology (such as SIFT matching features) is used to calculate the offset and correct the screen jitter. The coordinate offset compensation includes the following steps: a. Use image registration technology to correct the offset and jitter of the computer interlocking screen in real time; b. Use fixed feature points of the track or turnout (such as turnout tip position, track end point) as a reference to calculate coordinate offset and compensate S3. Color feature extraction: Generate a corresponding color mask based on the color range (red, white, blue, pink, yellow, etc.) defined by the track and switch status. Calculate the number of non-zero pixels in each color mask and determine the color status of the track and switch based on the pixel percentage. If the red mask contains more than 60% of the pixels, the track is considered occupied. The color mask generation includes the following steps: a. Create multiple sets of masks based on the preset HSV color range (e.g. red, white, blue, pink, yellow, green); preset different color HSV ranges based on the color characteristics of the track or turnout status, for example: Red: Hue∈[0,10]∪[170,180], Saturation∈[100,255], Value∈[100,255]; Blue: Hue∈[100,130], Saturation∈[100,255], Value∈[100,255]; White: Hue∈[0,180], Saturation∈[0,30], Value∈[200,255]; Yellow: Hue∈[20,30], Saturation∈[100,255], Value∈[200,255]; Green: Hue∈[50,70], Saturation∈[40,255], Value∈[40,255] Pink flash: Hue∈[140,170], Saturation∈[100,255], Value∈[100,255], simulates flash effect by changing frame rate (△V / △t>50); Convert the input RGB image to HSV color space to more accurately separate color information; filter the image pixel by pixel according to the predefined color range: mark the pixels that meet the color range as white, and mark the pixels that do not meet the color range as black (0), forming a binary mask image; For example: if the red state is detected, only the pixels in the image that meet the red HSV range are retained, and the pixels and the area are set to black to generate a red mask; b. Calculate the number of non-zero pixels in each mask group and compare the pixel percentages of each color mask. Select the color state corresponding to the largest percentage as the preliminary judgment result. Traverse the generated mask image and count the number of all white pixels (non-zero values). For example, if there are 1000 white pixels in the red mask, it means that the total number of pixels in the red area of ​​the image is 1000. Calculate the percentage of non-zero pixels in each color mask to the total non-zero pixels to determine the current state. Color ratio = ; Dynamic judgment of special states: Blocked state (pink flash): Determine whether the flash condition is met by the pixel change rate (△V / △t) of consecutive frames; for example: If the brightness value (V) of the pink pixel in a frame changes by more than a threshold value (such as 50) compared with the previous frame, it is determined to be a flash state S4. Shape feature extraction: The shape features of the track and turnout are extracted by contour detection technology. The contour of the track or turnout is extracted by the Canny edge detection algorithm, the geometric boundary of the track is identified, and the straight line feature and circular feature are detected by combining the Hough transform; the single-lock circular feature of the turnout is detected by combining the convex hull fitting method, the contour is fitted with a convex hull, and the geometric properties of the single-lock circular feature (such as circularity and area) are identified. For example, the circularity of the single-lock circular feature must meet ,in is the area, is the perimeter; the rectangularity must satisfy: width / height∈[0.9-1.1] to be considered a rectangle; The contour detection technique includes the following steps: a. Extract the outlines of tracks and switches using edge detection algorithms; b. Perform convex hull fitting on the contour to identify the geometric properties of the single-lock circular feature (such as circularity and area) S5. Layered feature fusion: Color features are combined with shape features to generate a comprehensive basis for judging track and switch status. For switch status, the occupancy status of the positioning track and the reverse track is combined to assist in judging the positioning, reverse and squeeze status of the switch. The judgment of tracks and switches in S5 is as follows: a. For track status: a1. If the color feature is blue, the track is considered to be idle; a2. If the color feature is red, it is determined that the track is occupied; a3. If the color feature is white and there is a direction arrow, it is determined that the route is locked; if there is no direction arrow, it is determined to be an abnormal state and a warning is issued; a4. If the color feature shows pink flashing, it is determined that the track is blocked; b. For turnout status: b1. If the turnout name and turnout tip position are displayed in green, it is considered to be positioned; b2. If the turnout name and turnout tip position are displayed in yellow, it is considered to be in reverse position; b3. If there is a green circle at the fork tip, it is considered a single lock; b4. If the turnout name is surrounded by a red box, it is considered blocked. The red pixel ratio is first detected using the red HSV mask. If the red pixel ratio exceeds the threshold, the rectangularity is also detected. If the rectangularity meets the requirement, it is considered blocked. If not, the next step is performed. b5. If the switch name is steady red and the switch tip is flashing red, it is considered a crowded switch; The turnout status judgment further includes the following steps: a. When both the positioning and anti-position tracks are idle, that is, the color feature at the track is displayed in blue, the idle state of the track is judged by the shape and color features: a1. Prioritize the red pixel ratio using the red HSV mask. If the red pixel ratio exceeds the threshold, the state is determined to be blocked. If the non-zero pixel ratio of the red mask is greater than 30%, the state is determined to be blocked. If the non-zero pixel ratio of the red mask is less than or equal to 30%, proceed to the next step. If the switch is not locked, the shape features of the green and yellow areas are extracted using contour detection technology to further determine the turnout's position and inversion status. The contours of the green or yellow areas are extracted using the Canny edge detection algorithm, and the convex hull is fitted to the contours to calculate the circularity. If the circularity is greater than 0.8, the turnout is determined to be in a single-lock state. If the single-lock feature is not detected, the position or inversion status is further determined based on the proportion of non-zero pixels in the green or yellow mask. b. When the positioning and anti-position orbital states are inconsistent: b1. Determine the initial state of the turnout based on the occupied status of the positioning and reverse tracks; if the positioning track is idle and the reverse track is occupied, the turnout is determined to be in reverse state; if the reverse track is idle and the positioning track is occupied, the turnout is determined to be in the positioning state; b2. Use contour detection combined with the convex hull fitting method to detect whether there is a circular feature of a single lock. Extract the contour of the green area and calculate the circularity. If a green circle with a circularity greater than 0.8 is present, it is determined to be in a single lock state. If there is no single lock state, proceed to the next step of judgment. b3. If there is no single lock feature, the final state of the turnout is determined by a comprehensive analysis of color and shape features. If the turnout name is surrounded by a red box, it is determined to be in a blocked state. If the turnout name is solid red and the turnout tip flashes red, it is determined to be in a squeezed state. If none of the above conditions are met, the position or reverse state is determined based on the proportion of non-zero pixels in the green or yellow mask. S6. Status output: Output the final status of the track and turnout based on the fused feature data, including idle, occupied, route locked, blocked, positioned, reversed, single locked, and squeezed turnout. The final status of the track or turnout is displayed in real time on the intelligent busy board in text, color, or graphic form. The status information is transmitted to the railway control system through a communication interface (such as CAN bus or Ethernet) for safety management; The status output includes the following steps: a. Display the final status of the track and turnout in real time on the intelligent busy board in any form of text, color or graphics; b. Transmit status information to the railway control system through the communication interface for track and switch safety management; if communication interruption or data loss is detected, an alarm is immediately triggered and recorded in the log database.

[0021] This embodiment provides an intelligent analysis system for track switch status based on hierarchical feature fusion, including: Image acquisition module, used to obtain video streams of the track or switch area, using a high-definition camera or industrial camera to obtain video streams of the track or switch area at a frame rate of 30fps; Image preprocessing module for performing image denoising, coordinate offset compensation, and color mask generation; Feature extraction module, used to extract color features (HSV mask) and shape features (contour, convex hull); Hierarchical feature fusion module, used to combine color features and shape features for state judgment; The status output module is used to output the final status information of the track or switch.

[0022] The hierarchical feature fusion module also includes: Track status judgment submodule, used to judge the track's idle, occupied, route locked or blocked status based on color characteristics; The turnout status judgment submodule is used to judge the turnout's positioning, reverse position, single lock, blockage or squeezed state by combining color features, shape features and track occupancy status.

[0023] In summary, this case solved the problems of poor dynamic adaptability and data silos in traditional turnout status judgment through a hierarchical and progressive logical architecture and multi-source data fusion technology, achieving a leap from "single perception" to "intelligent decision-making."

[0024] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent analysis method for track switch status based on intelligent busy board based on hierarchical feature fusion, characterized by: The following steps are involved: S1. Image preprocessing: De-noise the track and switch area images in the video stream, use erosion and dilation operations to remove noise in the signal light area, and optimize image quality through threshold adjustment in the HSV color space; S2. Coordinate offset compensation: According to the offset jitter characteristics of the microcomputer interlocking screen, coordinate correction is performed on the pre-processed image to adapt to the screen offset; S3, color feature extraction: Generate the corresponding color mask according to the color range defined by the track and switch status; Count the number of non-zero pixels in each color mask and determine the color status of the track and switch based on the pixel proportion; S4. Shape feature extraction: The shape features of the track and turnout are extracted using contour detection technology; the single-lock circular feature of the turnout is detected using the convex hull fitting method; S5. Layered feature fusion: Color features are combined with shape features to generate a comprehensive basis for judging track and switch status. For switch status, the occupancy status of the positioning track and the reverse track is combined to assist in judging the positioning, reverse and squeeze status of the switch. S6. Status output: Output the final status of the track and switch based on the fused feature data.

2. The intelligent analysis method for track switch status based on intelligent busy board based on hierarchical feature fusion according to claim 1 is characterized by: The color mask generation in S3 includes the following steps: a. Create multiple sets of masks based on the preset HSV color range; b. Calculate the number of non-zero pixels in each mask group, compare the pixel proportions of each color mask, and select the color state corresponding to the largest proportion as the preliminary judgment result.

3. The intelligent analysis method for intelligent busy board track switch status based on hierarchical feature fusion according to claim 2 is characterized by: The judgment of the track and turnout in S5 is as follows: a. For track status: a1. If the color feature is blue, the track is considered to be idle; a2. If the color feature is red, it is determined that the track is occupied; a3. If the color feature is white and there is a direction arrow, it is determined that the route is locked; if there is no direction arrow, it is determined to be an abnormal state and a warning is issued; a4. If the color feature shows pink flashing, it is determined that the track is blocked; b. For turnout status: b1. If the turnout name and turnout tip position are displayed in green, it is considered to be positioned; b2. If the turnout name and turnout tip position are displayed in yellow, it is considered to be in reverse position; b3. If there is a green circle at the fork tip, it is considered a single lock; b4. If the turnout name is surrounded by a red box, it is considered blocked; b5. If the switch name is steady red and the switch tip flashes red, it is considered a squeeze switch.

4. The intelligent analysis method for track switch status based on intelligent busy board based on hierarchical feature fusion according to claim 3 is characterized by: The switch state judgment in S5 further includes the following steps: a. When both the positioning and anti-position tracks are idle: a1. Prioritize detecting the red pixel ratio through the red HSV mask. If the red pixel ratio exceeds the threshold, it is judged as a blocked state. If the turnout is not blocked, the shape features of the green and yellow areas are extracted using contour detection technology to further determine the turnout's position and reverse status. b. When the positioning and anti-position orbital states are inconsistent: b1. Determine the initial state of the turnout based on the occupation status of the positioning and counter-position tracks; b2. Use contour detection technology combined with convex hull fitting to detect whether there is a circular feature of a single lock; b3. If there is no single lock feature, the final state of the turnout is determined by comprehensive analysis of color and shape features.

5. The intelligent analysis method for track switch status based on intelligent busy board based on hierarchical feature fusion according to claim 1 is characterized by: The image preprocessing in S1 further includes the following steps: a. Adjust the HSV color space for high-noise images and optimize the color threshold range through debugging; b. Perform local filtering on the noise points in the traffic light area to improve the accuracy of the color mask.

6. The intelligent analysis method for intelligent busy board track switch status based on hierarchical feature fusion according to claim 5 is characterized by: The coordinate offset compensation in S2 includes the following steps: a. Use image registration technology to correct the offset and jitter of the computer interlocking screen in real time; b. Use the fixed feature points of the track or switch as a reference to calculate the coordinate offset and compensate for it.

7. The intelligent analysis method for intelligent busy board track switch status based on hierarchical feature fusion according to claim 6 is characterized by: The contour detection technique in S4 includes the following steps: a. Extract the outlines of tracks and switches using edge detection algorithms; b. Perform convex hull fitting on the contour to identify the geometric properties of the single-lock circular feature.

8. The intelligent analysis method for intelligent busy board track switch status based on hierarchical feature fusion according to claim 7 is characterized by: The status output in S6 includes the following steps: a. Display the final status of the track and turnout in real time on the intelligent busy board in any form of text, color or graphics; b. Transmit status information to the railway control system through the communication interface for safe management of tracks and switches.

9. An intelligent analysis system for track switch states based on hierarchical feature fusion, based on the intelligent analysis method for track switch states based on hierarchical feature fusion according to any one of claims 1 to 8, characterized in that: include: Image acquisition module, used to obtain video streams of track or turnout areas; Image preprocessing module for performing image denoising, coordinate offset compensation, and color mask generation; Feature extraction module, used to extract color features and shape features; Hierarchical feature fusion module, used to combine color features and shape features for state judgment; The status output module is used to output the final status information of the track or switch.

10. The intelligent analysis system for track switch status based on intelligent bus board and hierarchical feature fusion according to claim 9 is characterized by: The hierarchical feature fusion module also includes: Track status judgment submodule, used to judge the track's idle, occupied, route locked or blocked status based on color characteristics; The turnout status judgment submodule is used to judge the turnout's positioning, reverse position, single lock, blockage or squeezed state by combining color features, shape features and track occupancy status.

Images