Safety Monitoring Method for Air Defense Sealed Partition Doors Using Photogrammetry

Abstract

This application provides a method for monitoring the safety of air-raid shelter doors using photogrammetry. A first and second reference target are set above the doorway, and a first and second measuring target are set on the side of the safety door. The first, second, and third reference targets are observed and imaged using an observation station. The images are processed, and the pixel coordinates of the first and second measuring targets are calculated using the first and second reference targets as references. The status of the safety door is determined based on the pixel coordinates and the orientation of their connection. The instrument's measurement data is connected to a cloud server via the internet and presented on a monitoring cloud platform. When the measurement data exceeds a set alarm threshold, an alarm mechanism is triggered. This method solves the problem of monitoring the status of air-raid shelter doors in subways.

Description

Technical Field

[0001] This invention relates to the field of subway safety monitoring, and in particular to a method for monitoring the safety of air-raid shelter sealed partition doors using photogrammetry. Background Technology

[0002] Subway civil defense engineering utilizes the existing structural strength and facilities of the subway system to enhance protection at key locations such as subway station entrances and exits, achieving civil defense requirements through specific peacetime-wartime conversion measures. Civil defense engineering is a crucial component of subway construction; in peacetime, the subway primarily operates for transportation, while in wartime it serves as a lifeline for personnel transfer and the transport of vital supplies.

[0003] After installation, the protective airtight partition doors of subway sections and lead-out lines may experience a decline in the performance of the door panel connectors due to their own weight or other external forces, such as human damage or continuous action of tunnel piston wind. This can lead to changes in the door panel's posture, such as the door panel falling (hanging corner) or the door panel shifting left or right (horizontally). If the door panel shifts severely, it may cause the door panel to intrude into the train operation area (track area), posing a great safety hazard to train operation.

[0004] For this type of monitoring, the instruments currently available are generally inclinometers and CCD laser displacement sensors. However, for objects that remain stationary year-round, these sensors suffer from problems such as data drift, short lifespan, and demanding installation requirements. Therefore, no more suitable monitoring method has been found. Summary of the Invention

[0005] This invention provides a method for monitoring the safety of air-raid shelter sealed partition doors using photogrammetry, which solves the problem of monitoring the status of air-raid shelter safety doors in subways.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a safety monitoring method for air-raid shelter sealed partition doors using photogrammetry.

[0007] A first reference target and a second reference target are set above the doorway, and a first measuring target and a second measuring target are set on the side of the safety door;

[0008] The first reference target, the second reference target, the first measurement target, and the second measurement target were observed and imaged by the observation station;

[0009] The image is processed, and the pixel coordinates of the first and second measurement targets are calculated based on the first and second reference targets.

[0010] The state of the safety gate is determined based on the pixel coordinates and connection posture of the first and second measuring targets.

[0011] The instrument's measurement data is connected to a cloud server via the Internet and presented on the monitoring cloud platform;

[0012] When the measured data exceeds the set alarm threshold, the alarm mechanism is triggered.

[0013] The preferred approach includes a target data image analysis method:

[0014] Adjust the line connecting the first and second reference targets to be horizontal, and adjust the line connecting the first and second measuring targets to be vertical;

[0015] The observation station was used to image the first reference target, the second reference target, the first measurement target, and the second measurement target in their initial state.

[0016] Take the midpoint of the line connecting the first and second reference targets on the image as the origin, the line connecting the first and second reference targets as the X-axis, and the axis perpendicular to the line connecting the first and second reference targets and passing through the origin as the Y-axis;

[0017] Draw a warning line parallel to the Y-axis at the point where the X-value of the doorway's outline on the image is maximized;

[0018] After debugging, the targets are photographed periodically, and the pixel coordinates of the first and second measurement targets are read. If the line connecting the first and second measurement targets intersects the warning line, it means that the security door has entered the security area.

[0019] The preferred solution includes a method for detecting security door intrusion:

[0020] During the installation and debugging of the targets, the distance from the warning line to the Y-axis S1 was measured, the distance from the first target to the upper edge of the safety door was L3, and the distance from the second target to the lower edge of the safety door was L4.

[0021] During periodic inspections, the distances of the first measurement target to the Y-axis (L1), the first measurement target to the X-axis (H1), the second measurement target to the Y-axis (L2), and the second measurement target to the X-axis (H2) are measured after imaging.

[0022] When either the first condition S1>=|L1-L3| or the second condition S1>=|L2-L4| is true, it indicates that the security door has intruded into the security area.

[0023] In a preferred embodiment, the target is provided with a first reflective part, a second reflective part and a third reflective part, the second reflective part being located above or below the first reflective part, and the third reflective part being located to the left or right of the first reflective part.

[0024] In a preferred embodiment, a fourth reflective part is also provided. The first, second, third, and fourth reflective parts are circular or regular polygonal. The first, second, third, and fourth reflective parts are arranged in a quadrilateral shape. The first and fourth reflective parts are arranged diagonally, as are the second and third reflective parts. The diameters or circumscribed circle diameters of the second and third reflective parts are the same or different. The diameters or circumscribed circle diameters of the second and third reflective parts are different from those of the first and fourth reflective parts.

[0025] In the preferred embodiment, the first reflective part, the second reflective part, the third reflective part, and the fourth reflective part are circular, and the first reflective part, the second reflective part, the third reflective part, and the fourth reflective part are internally tangent to the corners of the rectangular boundary.

[0026] The preferred approach includes a target-specific identification method:

[0027] Images were taken using an observation station;

[0028] Locate the bright areas of the image;

[0029] The bright areas of the image are converted into grayscale images. The base grayscale value of the dark pixel blocks is zero. The grayscale value of each pixel block is assigned according to the grayscale condition.

[0030] Once a bright area of ​​the image is located, the program iterates through the gray values ​​of that area to find features that match the four concentrated bright areas.

[0031] Find the boundaries of each bright area concentration region, calculate the diameter of each bright area concentration region, and fit a virtual circle. If the diameter and position distribution of each virtual circle are consistent with the target circle, it indicates that the bright area is the target image.

[0032] The preferred scheme includes a target center calculation method:

[0033] Imaging of the target;

[0034] The image is converted into a grayscale image, with the base grayscale value of the dark pixel block being zero, and grayscale values ​​are assigned to each pixel block according to the grayscale condition.

[0035] The program iterates through the gray values ​​of the region, finds the boundaries with non-zero gray values, fits a virtual rectangle, and the center of the virtual rectangle is the center of the target image.

[0036] The preferred solution includes a method for measuring the tilt of the safety door:

[0037] Read the pixel position coordinates of the first and second measurement targets, and the line connecting the original positions of the first and second measurement targets is denoted as the initial line. The distance of the initial line from the Y-axis is S2. The distance between the straight lines of the first and second measurement targets is denoted as the initial distance, S3.

[0038] The distance DX1 from the first measurement target to the initial line is DX1 = S2 - L1, and the distance DX2 from the second measurement target to the initial line is DX2 = L2 - S2;

[0039] The deflection angle a of the safety door is a = sin[|DX1| + |DX2|] / S3.

[0040] In a preferred solution, it includes a method for detecting the sinking and corner dropping of the safety door:

[0041] Read the pixel positions of the first measurement target and the second measurement target. The distance of the first measurement target relative to the X-axis is denoted as H of the first measurement target, and the distance of the second measurement target relative to the X-axis is denoted as H of the second measurement target. The sinking amount DH1 of the first measurement target is DH1 = H3 - H1, and the sinking amount DH2 of the second measurement target is DH2 = H4 - H2;

[0042] If DH1 > 0 and DH2 > 0 and H4 - H3 < H2 - H1, it indicates that the safety door has sunk and the corner has dropped.

[0043] The beneficial effects of the present invention are as follows: By setting a reference target, it avoids the reference drift caused by the displacement of the instruments in the observation station; the reference target is set above a relatively stable door opening structure, which is not easy to shift and deform, making the reference stable; the observation equipment can also borrow the existing monitoring equipment, which has a lower cost compared to the method of additional hardware layout such as inclination sensors, and does not require additional maintenance means; since there is no complex electrical device, the debugging and maintenance frequency is low, and the maintenance cost and labor cost are greatly reduced; the target has vector characteristics and can reflect its own attitude angle, making the detection more accurate. Brief Description of the Drawings

[0044] The present invention will be further described below with reference to the drawings and embodiments.

[0045] Figure 1 It is a top view schematic diagram of the present invention.

[0046] Figure 2 It is a front view schematic diagram of the present invention.

[0047] Figure 3 It is a schematic diagram of the opening and closing of the safety door of the present invention.

[0048] Figure 4 It is a diagram of the communication and alarm method of the present invention.

[0049] Figure 5 It is a measurement flow chart of the present invention.

[0050] Figure 6 It is a simplification of the relative position of the target of the present invention Figure 1 .

[0051] Figure 7 The target relative position is simplified in this invention. Figure 2 .

[0052] Figure 8 The target relative position is simplified in this invention. Figure 3 .

[0053] Figure 9 The target relative position is simplified in this invention. Figure 4 .

[0054] Figure 10 The target relative position is simplified in this invention. Figure 5 .

[0055] Figure 11 This is a schematic diagram of the target of the present invention.

[0056] Figure 12 This is the target imaging grayscale image of the present invention.

[0057] Figure 13 This is a schematic diagram of target imaging grayscale value fitting according to the present invention.

[0058] Figure 14 This is a target imaging centroid location diagram of the present invention.

[0059] Figure 15 This is the target imaging self-attitude detection diagram of the present invention.

[0060] Figure 16 This is a magnified image of the target in the distant view according to the present invention.

[0061] Figure 17 This is a composite image of the target distant view of the present invention.

[0062] Figure 18 This is a schematic diagram of the variable angle shooting of the present invention.

[0063] In the figure: First reference target 1; First reflective part 101; Second reflective part 102; Third reflective part 103; Fourth reflective part 104; Rectangular boundary 105; Second reference target 2; First measuring target 3; Second measuring target 4; Observation station 5; Safety door 6; Doorway 7. Detailed Implementation

[0064] Example 1:

[0065] like Figure 1-15 A method for safety monitoring of airtight partition doors in civil defense facilities using photogrammetry.

[0066] A first reference target 1 and a second reference target 2 or more are set above the doorway 7, and a first measuring target 3 and a second measuring target 4 or more are set on the side of the safety door 6;

[0067] The target uses a reflector, and the observation station is equipped with an infrared emitter. It uses reflective imaging and does not require a traditional light source.

[0068] The observation station 5 should ideally be located on one side of the door hinge to avoid obstructing the view when trains pass by. Since it is in a dark environment for a long time, thermal imaging camera equipment can be selected for the observation station 5. The first reference target 1, the second reference target 2, the first measurement target 3 and the second measurement target 4 can be observed and imaged through the observation station 5, and the four targets can be captured in the same image.

[0069] The image is processed, and the pixel coordinates of the first measurement target 3 and the second measurement target 4 are calculated based on the first reference target 1 and the second reference target 2.

[0070] The state of the safety door 6 is determined based on the pixel coordinates and connection posture of the first measuring target 3 and the second measuring target 4.

[0071] The instrument's measurement data is connected to a cloud server via the Internet and presented on the monitoring cloud platform;

[0072] When the measured data exceeds the set alarm threshold, the alarm mechanism is triggered.

[0073] Visual measurement is a photogrammetric method that calculates displacement changes by analyzing pixel changes at known points and the distance between the camera and these known points. The "known points" described by this method are called "targets." Targets can take various forms, including light spots created by LED light sources or specific feature points within the image. Both of these methods share a common problem: they cannot tolerate strong ambient light, as this can negatively impact the measurement results. For example, the measurement data may contain too many outliers, or the target may become completely unidentifiable.

[0074] Furthermore, in principle, the imaging of feature patterns depends on illumination. Traditional all-weather visual measurement techniques use luminous points as targets to increase the contrast between the target and surrounding objects. Moreover, changes in ambient light intensity can affect the measurement results. If the ambient light is too strong, causing the target imaging contrast to be insufficient, the measurement will become impossible.

[0075] From the perspective of energy consumption and construction deployment, targets using this method typically require a separate power supply and necessitate cable laying. The greater the measurement distance, the greater the workload. Furthermore, in some environments, cable laying may not be feasible, and the targets require a continuous working cycle of more than three months or even three years.

[0076] The preferred approach includes a target data image analysis method:

[0077] Adjust the line connecting the first reference target 1 and the second reference target 2 to be horizontal, and adjust the line connecting the first measuring target 3 and the second measuring target 4 to be vertical;

[0078] The first reference target 1, the second reference target 2, the first measurement target 3, and the second measurement target 4 in their initial states were imaged using observation station 5.

[0079] Take the midpoint of the line connecting the first reference target 1 and the second reference target 2 on the image as the origin, take the line connecting the first reference target 1 and the second reference target 2 as the X-axis, and take the axis perpendicular to the line connecting the first reference target 1 and the second reference target 2 and passing through the origin as the Y-axis;

[0080] Draw a warning line parallel to the Y-axis at the maximum X value point of the outline of doorway 7 on the image;

[0081] After debugging, the targets are photographed periodically, and the pixel coordinates of the first measuring target 3 and the second measuring target 4 are read. If the line connecting the first measuring target 3 and the second measuring target 4 intersects the warning line, it means that the security door 6 has entered the security area.

[0082] Since the first measuring target 3 and the second measuring target 4 are not necessarily installed against the top and bottom edges of the door, there is a certain distance between the targets and the door edge. If deformation occurs at this point... Figure 8 At the L3 mark, even though the first measurement target 3 did not intrude into the safe area, the actual door body had already intruded.

[0083] Therefore, the preferred solution includes a method for determining security door intrusion:

[0084] During the installation and debugging of the targets, the distance of the warning line from the Y-axis S1 was measured, the distance of the first measuring target 3 from the upper edge of the safety door 6 was L3, and the distance of the second measuring target 4 from the lower edge of the safety door 6 was L4.

[0085] During periodic inspections, after imaging, the distances of the first measuring target 3 to the Y-axis L1 and the first measuring target 3 to the X-axis H1 are measured, and the distances of the second measuring target 4 to the Y-axis L2 and the second measuring target 4 to the X-axis H2 are measured.

[0086] When either the first condition S1>=|L1-L3| or the second condition S1>=|L2-L4| is true, it indicates that security door 6 has intruded into the security area.

[0087] Target imaging uses reflective imaging from a distance, that is, emitting infrared light.

[0088] Since the observation device of this system may borrow existing equipment, the original monitoring location may be 10-100m away from the doorway section. For long-distance imaging, if there is only one reflector, the image captured from a distance is small, and the reflector itself is also small. In addition, due to the problem of glare during shooting, it is not easy to identify the center point of the image. If a large reflector is used, it is easy to be confused with single-point light sources such as lamps during shooting.

[0089] Therefore, in the preferred embodiment, the target is provided with a first reflective part 101, a second reflective part 102 and a third reflective part 103, the second reflective part 102 is provided above or below the first reflective part 101, and the third reflective part 103 is provided to the left or right of the first reflective part 101.

[0090] The three reflective parts are arranged in a way similar to coordinate axes, with a certain distance between them. The target is transformed into a vector graphic with a certain directionality.

[0091] In a preferred embodiment, a fourth reflective part 104 is also provided. The first reflective part 101, the second reflective part 102, the third reflective part 103, and the fourth reflective part 104 are circular or regular polygonal. The first reflective part 101, the second reflective part 102, the third reflective part 103, and the fourth reflective part 104 are arranged in a quadrilateral shape. The first reflective part 101 and the fourth reflective part 104 are arranged diagonally, and the second reflective part 102 and the third reflective part 103 are arranged diagonally. The diameters or circumscribed circle diameters of the second reflective part 102 and the third reflective part 103 are the same or different. The diameters or circumscribed circle diameters of the second reflective part 102 and the third reflective part 103 are different from those of the first reflective part 101 and the fourth reflective part 104.

[0092] The added fourth reflective part 104 further increases the target's specificity. In addition, the different diameters of each reflective part make it clearly distinguishable from ordinary lights or other patterns.

[0093] In a preferred embodiment, the first reflective part 101, the second reflective part 102, the third reflective part 103, and the fourth reflective part 104 are circular, and the first reflective part 101, the second reflective part 102, the third reflective part 103, and the fourth reflective part 104 are internally tangent to each corner of the rectangular boundary 105.

[0094] Any side or corner of the rectangular boundary 105 is equivalent to a coordinate axis, therefore the angle and attitude of each target are relatively easy to determine, such as... Figure 15 .

[0095] The preferred approach includes a target-specific identification method:

[0096] Images were taken using observation station 5;

[0097] Locate the bright areas of the image;

[0098] The bright areas of the image are converted into grayscale images. The base grayscale value of the dark pixel blocks is zero. The grayscale value of each pixel block is assigned according to the grayscale condition.

[0099] Once a bright area of ​​the image is located, the program iterates through the gray values ​​of that area to find features that match the four concentrated bright areas.

[0100] Find the boundaries of each bright area concentration region, calculate the diameter of each bright area concentration region, and fit a virtual circle. If the diameter and position distribution of each virtual circle are consistent with the target circle, it indicates that the bright area is the target image.

[0101] like Figure 16 For targets shot from a distance, due to the influence of pixel count and lens quality, there will be some missing pixels, making it inaccurate to directly confirm the center of the first reflective part 101.

[0102] Therefore, the preferred solution includes a target center calculation method:

[0103] Imaging of the target;

[0104] The image is converted into a grayscale image, with the base grayscale value of the dark pixel block being zero, and grayscale values ​​are assigned to each pixel block according to the grayscale condition.

[0105] The program iterates through the gray values ​​of the region, finds the boundaries with non-zero gray values, fits a virtual rectangle, and the center of the virtual rectangle is the center of the target image.

[0106] Since the target is a pattern composed of four circles, the size and proportion of which are known, the distance between the target and the camera can be calculated from the pixel size of the circles and the pixel distance between the circles when imaging within the field of view. Since multiple circles are actually involved in the calculation, the total graphic proportion of the target is increased, making it easier to identify, especially when shooting from a distance. Compared with a single graphic, the error is greatly reduced, reaching an error rate of <5%.

[0107] To further enhance the realism of distant images, multiple still photographs are taken consecutively during shooting, and then the images are composited. There are three processing methods:

[0108] The first type is a camera that does not have a gimbal.

[0109] Capture multiple images, or break down a video into multiple frames;

[0110] The second type involves a camera equipped with a horizontal gimbal.

[0111] The original image centers the four targets. After taking one or more photos, the machine is tilted left and right at a certain angle to take one or more photos of each target.

[0112] The third type is that the camera is equipped with a 360-degree pan-tilt head.

[0113] In the original imaging, the four targets are centered as a whole. After taking one or more photos, the machine swings left and right and up and down by a certain angle and then takes one or more photos respectively.

[0114] Due to camera shake during shooting and the vibration caused by the operation of the subway, there are tiny displacements at the pixel level in the actually taken photos. Therefore, there are differences even when taking each photo at the original position.

[0115] In addition, the lens has anisotropy, and generally the imaging in the central area is better than that in the edge area. For the shooting device with a pan-tilt head, this can be fully utilized. For example, Figure 18 , the shooting angle can be changed to change the lens area used for imaging.

[0116] For the above three situations, taking the image taken in the centered state as the reference, after aligning the targets of multiple images, the images are stacked and synthesized to remove noise and improve the resolution of the images.

[0117] The brightness area of the synthesized image is more accurate and nearly circular. For example, Figure 17 , the error rate is further reduced.

[0118] In the preferred solution, it includes a method for measuring the inclination of the safety door:

[0119] Read the pixel position coordinates of the first measurement target 3 and the second measurement target 4, and record the connection line of the original positions of the first measurement target 3 and the second measurement target 4 as the initial line. The distance between the initial line and the Y-axis is S2. Record the linear distance between the first measurement target 3 and the second measurement target 4 as the initial distance, denoted as S3.

[0120] The distance DX1 from the first measurement target 3 to the initial line = S2 - L1, and the distance DX2 from the second measurement target 4 to the initial line = L2 - S2.

[0121] The inclination angle a of the safety door = sin(|DX1| + |DX2|) / S3.

[0122] In the preferred solution, it includes a method for detecting the sinking and corner dropping of the safety door:

[0123] Read the pixel positions of the first measurement target 3 and the second measurement target 4. The distance of the first measurement target 3 relative to the X-axis is denoted as H of the first measurement target 3, and the distance of the second measurement target 4 relative to the X-axis is denoted as H of the second measurement target 4. The sinking amount DH1 of the first measurement target 3 = H3 - H1, and the sinking amount DH2 of the second measurement target 4 = H4 - H2.

[0124] If DH1 > 0 and DH2 > 0 and H4 - H3 < H2 - H1, it indicates that the safety door has sunk and the corner has dropped.

[0125] Example 2:

[0126] System components:

[0127] The device for monitoring the posture and safety of air defense doors using photographic technology mainly consists of three parts: measuring instruments, measuring targets, and reference targets.

[0128] The measuring instrument mainly consists of four parts: a photography module, a lighting module, a computer module, and a 4G network module.

[0129] The instrument can be set up within 10 to 100 meters in front of a sealed partition door to achieve the measurement purpose, but it is necessary to maintain visibility between the observation station and the sealed partition door. Select an appropriate measurement distance based on the site environment.

[0130] Since the door is always open, the measuring targets are attached to the side of the sealed partition door using adhesive, along its thickness. The targets are weather-resistant paper patterns made of a highly reflective material, projected by the instrument as infrared invisible light. Available sizes are 10cm*10cm, 8cm*8cm, and 5cm*5cm, all with a thickness of 1mm. The appropriate target size will be selected based on the door's thickness. Two targets are placed on the side of the sealed partition door, one above the other. Figure 2-3 .

[0131] The reference point set consists of two targets, which are directly affixed near the gate. These two reference point sets effectively solve the problem of slow data drift caused by changes in the observation station's attitude due to frequent train movement, ensuring the reliability of long-term data trends.

[0132] Communication and alarm methods:

[0133] The measurement data is linked through 4G internet, cloud control, and on-site alarm devices. The instrument enables long-term real-time monitoring of the air-raid shelter doors. Data is transmitted to a cloud server via 4G network, and the subway operations department accesses the measurement data controlled by the cloud server in real time via the internet or a dedicated network. When data exceeds a threshold, management personnel are notified via SMS and platform interface notifications. After emergency safety confirmation, the platform activates the on-site alarm device, and the train takes emergency braking or delays upon receiving the alarm.

[0134] Technical advantages:

[0135] 1. High precision: The photographic technology can achieve a measurement accuracy of less than ±0.1mm for close-up scenes, and the reliability is high when calculating indicators such as tilt and settlement through the displacement of the target.

[0136] 2. Non-contact measurement: The target itself can be directly adhered and will not affect the opening and closing of the door. No power supply is required, and there are no sources of interference.

[0137] 3. Timeliness and accuracy of alarms: Instrument measurement data is connected to cloud services via 4G internet and presented on the monitoring cloud platform. In addition to real-time and historical data display, the platform features instrument status display, alarm push notifications, and alarm threshold settings. When the measurement data exceeds the set alarm threshold, the alarm mechanism is triggered. This is manifested as follows:

[0138] (1) The platform homepage displays alarm prompts;

[0139] (2) Push notifications via SMS and WeChat service account;

[0140] (3) Relevant personnel confirm the alarm by viewing real-time video surveillance.

[0141] Therefore, when the door displacement exceeds the limit, the relevant personnel of the operator can be notified in a timely manner, and the train conductor can be informed promptly.

[0142] 4. Reduce the frequency of periodic testing of airtight partition doors. Due to the high safety requirements of subway operations, even when the door is in a completely safe state, the protective airtight partition doors still require periodic inspection and testing. Automated real-time attitude monitoring can guide periodic inspections, significantly reducing maintenance costs. Real-time data trends provide precise guidance for testing.

[0143] 5. Scientific and reasonable safety evaluation: The measurement data is displayed in real time, reflecting the real-time deformation of the door. Massive amounts of data will be analyzed to determine the door's lifespan and provide guidance on factors affecting door safety.

[0144] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for safety monitoring of air-raid shelter sealed partition doors using photogrammetry, characterized in that: A first reference target (1) and a second reference target (2) are set above the doorway (7), and a first measuring target (3) and a second measuring target (4) are set on the side of the safety door (6). The first reference target (1), the second reference target (2), the first measurement target (3), and the second measurement target (4) are observed and imaged through the observation station (5); The image is processed, and the pixel coordinates of the first measurement target (3) and the second measurement target (4) are calculated based on the first reference target (1) and the second reference target (2). The state of the safety gate (6) is determined based on the pixel coordinates and connection posture of the first measuring target (3) and the second measuring target (4); The instrument's measurement data is connected to a cloud server via the Internet and presented on the monitoring cloud platform; When the measured data exceeds the set alarm threshold, the alarm mechanism is triggered; Including target data image analysis methods: Adjust the line connecting the first reference target (1) and the second reference target (2) to be horizontal, and adjust the line connecting the first measuring target (3) and the second measuring target (4) to be vertical; Using the observation station (5), images are taken of the first reference target (1), the second reference target (2), the first measurement target (3), and the second measurement target (4) in their initial states; Take the midpoint of the line connecting the first reference target (1) and the second reference target (2) on the image as the origin, take the line connecting the first reference target (1) and the second reference target (2) as the X-axis, and take the axis perpendicular to the line connecting the first reference target (1) and the second reference target (2) and passing through the origin as the Y-axis; At the point where the X-value of the outline on the image is at the doorway (7), draw a warning line parallel to the Y-axis; After debugging, the targets are photographed periodically, and the pixel coordinates of the first measurement target (3) and the second measurement target (4) are read. If the line connecting the first measurement target (3) and the second measurement target (4) intersects with the warning line, it means that the security door (6) has entered the security area.

2. The method for safety monitoring of air-raid shelter sealed partition doors using photogrammetry as described in claim 1, characterized in that: Including methods for detecting security door intrusion: During the installation and debugging of the targets, the distance of the warning line from the Y-axis S1 is measured. The distance of the first measuring target (3) from the upper edge of the safety door (6) is L3, and the distance of the second measuring target (4) from the lower edge of the safety door (6) is L4. During periodic testing, after imaging, the distance of the first measuring target (3) from the Y-axis L1, the distance of the first measuring target (3) from the X-axis H1, the distance of the second measuring target (4) from the Y-axis L2, and the distance of the second measuring target (4) from the X-axis H2 are measured. When either the first condition S1>=|L1-L3| or the second condition S1>=|L2-L4| is met, it indicates that the security door (6) has intruded into the security area.

3. The method for safety monitoring of air-raid shelter sealed partition doors using photogrammetry as described in claim 2, characterized in that: The target is provided with a first reflective part (101), a second reflective part (102) and a third reflective part (103). The second reflective part (102) is located above or below the first reflective part (101), and the third reflective part (103) is located to the left or right of the first reflective part (101).

4. The method for safety monitoring of air-raid shelter sealed partition doors using photogrammetry as described in claim 3, characterized in that: A fourth reflective part (104) is also provided. The first reflective part (101), the second reflective part (102), the third reflective part (103) and the fourth reflective part (104) are circular or regular polygonal. The first reflective part (101), the second reflective part (102), the third reflective part (103) and the fourth reflective part (104) are arranged in a quadrilateral shape. The first reflective part (101) and the fourth reflective part (104) are arranged diagonally. The second reflective part (102) and the third reflective part (103) are arranged diagonally. The diameter or circumscribed circle diameter of the second reflective part (102) and the third reflective part (103) is the same or different. The diameter or circumscribed circle diameter of the second reflective part (102) and the third reflective part (103) is different from that of the first reflective part (101) and the fourth reflective part (104).

5. The method for safety monitoring of air-raid shelter sealed partition doors using photogrammetry as described in claim 4, characterized in that: The first reflective part (101), the second reflective part (102), the third reflective part (103) and the fourth reflective part (104) are circular, and the first reflective part (101), the second reflective part (102), the third reflective part (103) and the fourth reflective part (104) are internally tangent to the corners of the rectangular boundary (105).

6. The method for safety monitoring of air-raid shelter sealed partition doors using photogrammetry as described in claim 5, characterized in that: Including target-specific identification methods: Images were taken using the observation station (5); Locate the bright areas of the image; The bright areas of the image are converted into grayscale images. The base grayscale value of the dark pixel blocks is zero. The grayscale value of each pixel block is assigned according to the grayscale condition. Once a bright area of ​​the image is located, the program iterates through the gray values ​​of that area to find features that match the four concentrated bright areas. Find the boundaries of each bright area concentration region, calculate the diameter of each bright area concentration region, and fit a virtual circle. If the diameter and position distribution of each virtual circle are consistent with the target circle, it indicates that the bright area is the target image.

7. The method for safety monitoring of air-raid shelter sealed partition doors using photogrammetry as described in claim 6, characterized in that: Including target center calculation methods: Imaging of the target; The image is converted into a grayscale image, with the base grayscale value of the dark pixel block being zero, and grayscale values ​​are assigned to each pixel block according to the grayscale condition. The program iterates through the gray values ​​of the region, finds the boundaries with non-zero gray values, fits a virtual rectangle, and the center of the virtual rectangle is the center of the target image.

8. The method for safety monitoring of air-raid shelter sealed partition doors using photogrammetry as described in claim 2, characterized in that: Including methods for measuring the tilt of safety doors: Read the pixel position coordinates of the first measurement target (3) and the second measurement target (4), and obtain the line connecting the original positions of the first measurement target (3) and the second measurement target (4) as the initial line. The distance of the initial line from the Y axis is S2. The distance between the straight lines of the first measurement target (3) and the second measurement target (4) is recorded as the initial distance, denoted as S3. The distance between the first measuring target (3) and the initial line is DX1=S2-L1, and the distance between the second measuring target (4) and the initial line is DX2=L2-S2; The angle of deflection of the safety door is a = sin[|DX1|+|DX2|] / S3.

9. The method for safety monitoring of air-raid shelter sealed partition doors using photogrammetry as described in claim 2, characterized in that: Including methods for detecting safety door sagging and corner damage: Read the pixel positions of the first measurement target (3) and the second measurement target (4). The distance of the first measurement target (3) relative to the X-axis is denoted as H_first measurement target (3), and the distance of the second measurement target (4) relative to the X-axis is denoted as H_second measurement target (4). The subsidence of the first measurement target (3) is DH1 = H3 - H1, and the subsidence of the second measurement target (4) is DH2 = H4 - H2; If DH1 > 0 and DH2 > 0 and H4 - H3 < H2 - H1, it indicates that the safety door has subsided and chipped off.

Images