Industrial door anti-pinch safety early warning method based on video stream moving target detection

Abstract

The present application belongs to the technical field of industrial automation and safety monitoring, and particularly relates to an industrial door anti-pinch safety early warning method based on video stream motion target detection, comprising the following steps: S1, collecting real-time video stream of an industrial door area, pre-processing and feature tracking the real-time video stream, mapping image coordinates to a world coordinate system, and obtaining real-time position parameters and motion speed parameters of a motion target; S2, calculating a radial approaching equivalent threat speed index according to the motion speed parameters of the motion target and the direction relative to the door body, the radial approaching equivalent threat speed index being used to represent the equivalent threat speed of the motion target sprinting towards the door body. The present application realizes differentiated and accurate response to high-speed forklifts and low-speed pedestrians by establishing a multi-dimensional dynamic collision risk model, and maximizes the passing efficiency of the industrial door under the premise of ensuring safety.

Description

Technical Field

[0001] This invention belongs to the field of industrial automation and safety monitoring technology, specifically relating to an industrial door anti-pinch safety early warning method based on video stream moving target detection. Background Technology

[0002] In modern industrial production and logistics warehousing environments, industrial high-speed doors serve as crucial nodes connecting different workshops, and their operational efficiency and safety are paramount. Industrial doors require frequent and rapid opening and closing to ensure smooth logistics, while simultaneously preventing injuries to personnel or collisions with vehicles during the closing process.

[0003] Current industrial door anti-pinch security primarily relies on photoelectric sensors or inductive loop detectors. With technological advancements, video-based area intrusion detection is also being applied. Existing video detection solutions typically define a fixed rectangular region of interest in an image and use background modeling algorithms or frame differencing to detect moving pixels within that region. This detection method is essentially a binary judgment logic: as long as an object moves within the region, regardless of its direction or speed, the system outputs a blocking signal, triggering the door to stop abruptly or open in the reverse direction.

[0004] However, in real-world logistics scenarios, existing technologies exhibit significant logical flaws. First, they fail to differentiate risk levels. Targets passing by a door laterally without entering, or those still far away and moving slowly, often do not pose an immediate collision risk. However, frequent false alarms from existing technologies can cause industrial doors to repeatedly open or fail to close, severely impacting the temperature-controlled clean environment of the workshop and logistics efficiency. Second, they lack dynamic spatiotemporal matching. A high-speed forklift and a slow-moving worker pose drastically different threats to the door at the same distance. Existing technologies use fixed distance thresholds, failing to adapt to the differentiated needs of forklifts stopping early and pedestrians stopping later, resulting in a trade-off between safety and efficiency. Summary of the Invention

[0005] This invention provides an industrial door anti-pinch safety early warning method based on video stream moving target detection, in order to solve the technical problems that existing industrial door safety systems cannot effectively distinguish the collision risk level of moving targets, and lack spatiotemporal dynamic matching of high-speed and low-speed targets.

[0006] This invention provides an industrial door anti-pinch safety early warning method based on video stream moving target detection, comprising the following steps: S1: Acquire real-time video stream from the industrial gate area, preprocess and track features of the real-time video stream, map image coordinates to the world coordinate system, and obtain real-time position and velocity parameters of the moving target. S2, Calculate the radial approximation equivalent threat velocity index based on the moving target's speed parameters and its direction relative to the door. The radial approximation equivalent threat velocity index is used to characterize the equivalent threat velocity of the moving target rushing towards the door. S3, obtain the current operating status of the gate, combine the radial approximation equivalent threat velocity index and real-time position parameters, calculate the spatiotemporal pressure coupling coefficient, the spatiotemporal pressure coupling coefficient is used to characterize the resource competition relationship between the gate's rated total closing time and the time required for the moving target to reach the gate; S4 determines the safety intervention probability based on the spatiotemporal pressure coupling coefficient, and outputs the corresponding control signal to the industrial door controller according to the numerical range of the safety intervention probability to perform safety maintenance, early warning deceleration or emergency braking operations.

[0007] The benefits are as follows: By constructing a radial approximation equivalent threat velocity index coupled with a spatiotemporal pressure coefficient, this invention transforms the visual information collected by the camera into a quantitative assessment of the competition for impact energy and time resources. This mechanism endows the control system with human-like perception and judgment capabilities: when faced with a high-speed forklift, the system can detect high-kinetic-energy threats from a greater distance and issue a warning and brake accordingly; while when faced with pedestrians approaching at low speeds or objects passing parallel to it, the system can tolerate their movement within a relatively close distance without triggering false alarms. Ultimately, this invention finds the optimal dynamic balance between absolute safety by eliminating the risk of physical collisions and efficient passage by reducing ineffective door openings, significantly improving the level of intelligent management in industrial sites.

[0008] Furthermore, the formula for calculating the radial approximation equivalent threat velocity index is as follows:

[0009] In the formula, To approximate the equivalent threat velocity index radially, The magnitude of the current instantaneous velocity of the moving target. For direction coefficients, As the reference velocity constant, It is an exponential function with the natural constant e as its base.

[0010] The advantages are as follows: This invention employs an exponentially growing function model to calculate the radially approximating equivalent threat velocity index. Utilizing the sensitivity of the exponential function to numerical changes, it simulates the kinetic energy law in physics where destructive force is proportional to the square of velocity. Compared to linear velocity monitoring, this calculation method non-linearly amplifies the system's threat perception of high-speed moving targets, ensuring that when high-speed moving heavy equipment is detected, the threat index quickly exceeds the threshold, triggering an earlier and more decisive braking response and effectively mitigating the collision risk caused by inertia.

[0011] Furthermore, the direction coefficient is the larger of the cosine of the angle between the world coordinate velocity vector of the moving target and the normal direction of the gate and zero; when the angle is less than 90 degrees, the direction coefficient is positive, indicating that the target is approaching the gate; when the angle is greater than or equal to 90 degrees, the direction coefficient is zero, indicating that the target is moving parallel to or away from the gate.

[0012] The effect is as follows: This invention achieves vectorized analysis of the trajectory of moving targets by introducing a directional coefficient to weight the velocity vector. This technique can accurately eliminate lateral interference commonly found in industrial scenarios. Specifically, for objects that merely pass laterally through the doorway without entering the doorway, or objects moving away from the door, the system automatically reduces their risk weight to zero. This solves the false alarm problem of traditional area intrusion detection technology, which alarms whenever there is a pixel change within the area, ensuring that industrial doors only react to entry behaviors with genuine collision intent.

[0013] Furthermore, the formula for calculating the spatiotemporal pressure coupling coefficient is as follows:

[0014] In the formula, The spatiotemporal pressure coupling coefficient. The rated total height of the door. The rated closing speed set for the door. This represents the real-time vertical distance between the moving target and the door plane. To achieve an equivalent safe approximation of the velocity constant, The overall system response delay time. To approximate the equivalent threat velocity index radially, The rated total closing time of the door. The target is the equivalent safe arrival time threshold.

[0015] The effect is as follows: This invention constructs a spatiotemporal pressure coupling coefficient, placing the two dynamic variables—the remaining time required for the door to close and the time required for the target to reach the impact point—into the same mathematical model for game-theoretic comparison. This is essentially calculating the width of the escape time window, rather than simply comparing physical distances. This design allows the system to dynamically adjust its safety strategy based on the door's current operating height, ensuring that at the critical moment when the door is about to close completely, any slight approach will trigger a high level of pressure, thus preventing accidents caused by people being trapped at the last moment.

[0016] Furthermore, the equivalent safety approach velocity constant is used to prevent the denominator from becoming zero when the radial approach equivalent threat velocity exponent is zero, and characterizes the minimum safety buffer threshold for stationary targets.

[0017] Furthermore, the formula for calculating the probability of security intervention is as follows:

[0018] In the formula, To ensure the probability of safe intervention, This is the risk critical threshold constant. Sensitivity index This is the spatiotemporal pressure coupling coefficient.

[0019] Furthermore, the sensitivity index is used to adjust the steepness of the response of the security intervention probability to changes in the spatiotemporal pressure coupling coefficient; the risk critical threshold constant is used to define the critical equilibrium point when the door closing time is equal to the target impact time.

[0020] Furthermore, preprocessing and feature tracking are performed on the real-time video stream, including: Distortion correction is performed on the captured video frames to restore the true geometric proportions of the scene; Gaussian filtering is applied to the corrected video frames to reduce interference from ambient light and sensor noise. The sparse optical flow algorithm is used to track feature corner points in video frames, and the pixel velocity of the moving target in the image coordinate system is obtained by calculating the displacement of the feature corner points between consecutive frames.

[0021] Furthermore, mapping the image coordinates to the world coordinate system includes: During the system initialization phase, the camera's intrinsic and extrinsic parameter matrices are determined using a calibration method, and a homography matrix mapping relationship between pixel coordinates and ground world coordinates is constructed. Based on the homography matrix, the pixel displacement of the moving target on the image plane is converted into the instantaneous movement speed on the real world ground in real time; Based on the homography matrix, calculate the vertical distance between the center of the feature point of the moving target and the plane where the door is located.

[0022] Furthermore, the corresponding control signal is output to the industrial door controller, including: A first security probability threshold and a second security probability threshold are preset, wherein the second security probability threshold is greater than the first security probability threshold; When the probability of safety intervention is less than the first safety probability threshold, it is determined to be a safe zone, and a signal to maintain the normal closure of the door is output. When the probability of safety intervention is greater than or equal to the first safety probability threshold and less than the second safety probability threshold, it is determined to be a warning interval, and a signal is output to trigger the warning light to flash and control the door to slow down and close. When the probability of safety intervention is greater than or equal to the second safety probability threshold, it is determined to be an emergency stop zone, and a signal to cut off the downward movement of the door and control the door to open in the reverse direction is output.

[0023] The beneficial effects are as follows: The core innovation of this invention lies in breaking the traditional binary opposition logic of immediate stop upon the presence of an object, and introducing the concepts of kinetic energy and spatiotemporal game theory from physics. By constructing a radial approach equivalent threat velocity index and a spatiotemporal pressure coupling coefficient, the system can, like the human brain, comprehensively judge the target's sprint trend and remaining opportunities. This gives the system dynamic adaptability; that is, for a forklift approaching at high speed, the system will intervene at a greater distance; while for a pedestrian approaching at low speed, the system allows them to move within a closer distance without triggering a false alarm. This differentiated processing logic fundamentally solves the long-standing contradiction between safety and efficiency in industrial scenarios, significantly improving the level of intelligent management in industrial sites. Attached Figure Description

[0024] Figure 1 This is a flowchart of the industrial door anti-pinch safety early warning method based on video stream moving target detection in this invention.

[0025] Figure 2 This is a comparison graph of the existing technology and the security response mechanism of this invention.

[0026] Figure 3 This is a dynamic risk differentiation analysis diagram based on velocity vector in this invention.

[0027] Figure 4 This is a distribution map of the three-dimensional spatiotemporal pressure risk field in this invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] An embodiment of the industrial door anti-pinch safety early warning method based on video stream moving target detection provided by the present invention: like Figure 1 As shown, the industrial door anti-pinch safety early warning method based on video stream moving target detection includes the following steps: S1 acquires real-time video streams from the industrial gate area, preprocesses and tracks the features of the real-time video streams, maps the image coordinates to the world coordinate system, and obtains the real-time position parameters and motion speed parameters of the moving targets.

[0030] This step forms the foundation of the entire system's perception capabilities. The system acquires real-time video streams using a top-down wide-angle industrial camera mounted on the top of the industrial door. To eliminate lens distortion and noise, distortion correction is first performed on the video frames. For example, geometric transformations are applied to the image using the camera's distortion coefficients to restore the true scene proportions. Subsequently, Gaussian filtering is performed for noise reduction. The image is smoothed using a Gaussian kernel function to remove random noise caused by insufficient lighting or sensor thermal noise, thereby improving image quality.

[0031] Next, the sparse optical flow algorithm is used to track the feature corner points in the image. Specifically, the algorithm detects corner points in the first frame of the image and calculates the displacement vectors of these corner points in subsequent frames through the optical flow constraint equation, thereby obtaining the pixel velocity of the moving target in the image coordinate system.

[0032] To perform physical calculations, image data must be mapped to a real-world coordinate system. During the installation phase, the camera's intrinsic and extrinsic parameter matrices are determined using Zhang Zhengyou's calibration method, and pixel coordinates are established. ground world coordinates The homography matrix mapping relationship is established. Based on this mapping relationship, the system calculates the following parameters in real time: (Target velocity): Calculate the pixel displacement of the target feature points between consecutive frames, and combine the frame rate and homography matrix to solve for the instantaneous movement velocity of the target on the real world ground.

[0033] (Target Distance): Calculates the vertical distance between the center of the target feature point and the plane where the door is located.

[0034] (Motion Angle): Calculates the angle between the target's world coordinate velocity vector and the direction of the door's normal.

[0035] Example: Assuming the camera is installed at a height of 5 meters, it detects a forklift moving downwards at the center of the image. The image processing module first performs distortion correction using bilinear interpolation. Next, an optical flow algorithm detects feature points on the forklift's forks and calculates its movement speed in the image as 10 pixels per frame. Combining the calibrated homography matrix and a frame rate of 30 FPS, the system calculates its actual ground speed. The speed is 3.5 meters per second, and the distance from the gate is... It is 8 meters long and the direction of movement is directly facing the gate, that is .

[0036] By acquiring video streams and mapping coordinates, two-dimensional image information can be transformed into motion parameters in the real physical world, providing reliable data support for subsequent precise physical calculations.

[0037] S2, calculate the radial approximation equivalent threat velocity index based on the moving target's speed parameters and its direction relative to the door. The radial approximation equivalent threat velocity index is used to characterize the equivalent threat velocity of the moving target rushing towards the door.

[0038] This step aims to separate ineffective and low-threat motions. Considering the square-law relationship between kinetic energy and velocity, and the severe consequences of high-speed impacts in industrial doorway scenarios, this invention employs exponential growth logic to amplify the threat value of high-speed targets. The formula for calculating the radial approximation equivalent threat velocity exponent is:

[0039] In the formula, To approximate the equivalent threat velocity index radially, The magnitude of the current instantaneous velocity of the moving target. The reference speed constant is set to the speed limit of forklifts in the factory, such as 3.0 m / s, and is used to normalize the exponential term.

[0040] The direction coefficient has a value of ,in Let be the angle between the target's direction of motion and the normal to the door. When hour, It is positive, and the more positive it is, the larger the value of the gate body; when hour, Setting it to 0 ensures that objects leaving the door will not generate a risk value.

[0041] Example: Assuming a reference speed m / s; Scenario A, Low Risk: A worker with Walking towards the door at a speed of m / s, that is Calculation process: .

[0042] Scenario B, High Risk: A forklift... A velocity of m / s is rushing towards the door; Calculation process: .

[0043] As can be seen, although the forklift's speed is only four times that of a pedestrian, its The index is more than 10 times that of a pedestrian, which makes the system extremely sensitive to high-speed targets, consistent with the physical law that kinetic energy destructive force increases non-linearly with speed.

[0044] By constructing a radial approximation equivalent threat velocity index, the magnitude and direction of the target's velocity can be coupled, and the threat value of high-speed targets can be amplified using exponential growth logic. Invalid targets with lateral movement can be filtered out, accurately reflecting the potential impact force of the target on the door.

[0045] S3: Obtain the current operating status of the gate, and calculate the spatiotemporal pressure coupling coefficient by combining the radial approximation equivalent threat velocity index and real-time position parameters. The spatiotemporal pressure coupling coefficient is used to characterize the resource competition relationship between the gate's rated total closing time and the time required for the moving target to reach the gate.

[0046] The core logic of this step is to compare how long it will take for the door to close with how long it will take for the car to crash into it. The formula for calculating the spatiotemporal pressure coupling coefficient is:

[0047] In the formula, The rated total closing time of the door. To achieve the target equivalent safe arrival time threshold, This is the spatiotemporal pressure coupling coefficient; the larger the value, the higher the collision risk. The rated total height of the door is read in real time by the door operator's encoder. The rated closing speed set for the door is a known fixed parameter. This represents the real-time vertical distance between the target and the plane of the door. It is the equivalent safe approach velocity constant for stationary targets, used to prevent the denominator from being zero when the radial approach equivalent threat velocity exponent is zero, and characterizes the minimum safe buffer threshold for stationary targets. This refers to the overall system response latency, including image processing time and mechanical braking delay.

[0048] Example: Assuming the door's rated total height m, closing speed m / s, response latency s, Then the rated total closing time of the door. Second.

[0049] Continuing from scene B in S2, Assuming distance m; Target equivalent safe arrival time threshold Seconds; Calculate the spatiotemporal pressure coupling coefficient: .

[0050] at this time A value much greater than 1 indicates that it takes 6 seconds for the door to close, while the danger could occur in just 0.72 seconds, posing an extremely high risk that requires immediate action.

[0051] If we continue with scenario A in S2 Assuming the distance is the same m; Target equivalent safe arrival time threshold Seconds; Calculate the spatiotemporal pressure coupling coefficient: .

[0052] at this time A value close to 1 indicates that although there is a risk, it is still on the verge of being controllable, and the system may choose to issue a warning and slow down rather than stop abruptly.

[0053] By constructing a spatiotemporal pressure coupling coefficient, the dynamic processes of door closing and target impact can be analyzed using a game theory approach. Instead of relying solely on distance judgment, the risk is assessed based on the scarcity of time resources, thus improving the scientific rigor of the judgment.

[0054] S4 determines the safety intervention probability based on the spatiotemporal pressure coupling coefficient, and outputs the corresponding control signal to the industrial door controller according to the numerical range of the safety intervention probability to perform safety maintenance, early warning deceleration or emergency braking operations.

[0055] This is a theoretical value that needs to be converted into a braking probability signal that can be executed by the PLC controller. The formula for calculating the safety intervention probability is:

[0056] In the formula, The probability of safe intervention is 0.0-1.0. This is a risk threshold constant, for example, 1.0, representing the critical point when the closing time is equal to the impact time. For example, the sensitivity index is taken as an integer, such as 4 or 6. The larger the value, the steeper the response curve, and the more decisive the system's response to risks of crossing the threshold.

[0057] The system according to Implement hierarchical control: safe range, for example If the risk is determined to be no risk or low risk, a safety maintenance signal is output to control the door to maintain its normal closing action.

[0058] Warning range, for example If a potential risk is identified, a warning deceleration signal is output, triggering a flashing yellow warning light, and the frequency converter is controlled to halve the door's closing speed. This can actively extend... Thus reducing This is valuable, allowing for more observation time.

[0059] Emergency stop zone, for example If a collision is detected, an emergency braking signal is output. The PLC immediately cuts off the downlink signal, triggers an emergency stop, and automatically reverses to the upper limit position, while simultaneously triggering a red audible and visual alarm.

[0060] Example: Continuing from scene B in S3, ;set up Calculate the probability: ;because The system determines that it has entered the emergency stop zone and immediately triggers the emergency stop reverse direction.

[0061] Continuing from scenario A in S3, Calculate the probability: .

[0062] because When the system enters the warning zone, the gate slows down and a yellow light illuminates. If the worker continues to approach, the risk value will continue to rise, eventually triggering an emergency stop. If the worker stops, the risk value decreases, and the gate closes normally.

[0063] By calculating the probability of safety intervention and implementing tiered control, it is possible to achieve the control effect of no disturbance in normal times and a complete stop in times of danger, which ensures both traffic efficiency under low risk and timely response under high risk.

[0064] To more intuitively demonstrate the effects of this invention, please refer to... Figure 2 , Figure 3 and Figure 4 .

[0065] Figure 2 This graph compares the safety response mechanism of the present invention with that of existing technologies. It is a two-dimensional curve comparison, with the horizontal axis representing the distance from the target to the door and the vertical axis representing the probability of safety intervention. The curve of the existing technology only jumps to 1 instantaneously when the distance is less than 2.5 meters, showing a significant lag. The curve of the present invention presents a smooth S-shaped curve, starting to rise at a distance of 8 meters, indicating early intervention. This clearly demonstrates the advantages of the present invention, which achieves early prediction and soft intervention through algorithms.

[0066] Figure 3 This demonstrates the difference in risk values ​​for targets at different speeds at the same distance. High-speed targets exceed the risk threshold at a distance of 7 meters. Low-speed targets only exceed the risk threshold when they are close to the gate. The filled area between the two curves is marked as a safe passage zone, vividly illustrating that this solution effectively improves the passage efficiency for low-speed targets, preventing accidental blocking of pedestrians in the name of vehicle prevention.

[0067] Figure 4 This is a three-dimensional spatiotemporal pressure risk field distribution map, a three-dimensional surface plot with three coordinate axes representing distance, approach velocity, and spatiotemporal pressure coupling coefficient. The surface shows a smooth transition from a safe zone at a lower level to a dangerous zone at a higher level. The plot visually demonstrates that the risk coefficient not only increases with decreasing distance but also exhibits a non-linear, sharp increase with increasing velocity, perfectly visualizing the core mathematical logic of the two formulas: the radial approach equivalent threat velocity index and the spatiotemporal pressure coupling coefficient.

[0068] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A safety early warning method for preventing pinching of industrial doors based on moving target detection in video streams, characterized in that, Includes the following steps: S1: Acquire real-time video stream from the industrial gate area, preprocess and track features of the real-time video stream, map image coordinates to the world coordinate system, and obtain real-time position and velocity parameters of the moving target. S2, based on the moving target's velocity parameters and its direction relative to the door, calculate the radial approximation equivalent threat velocity index. The radial approximation equivalent threat velocity index characterizes the equivalent threat velocity of the moving target rushing towards the door. for: The magnitude of the current instantaneous velocity of the moving target. For direction coefficients, As the reference velocity constant, It is an exponential function with the natural constant e as its base; S3, obtain the current operating status of the gate, combine the radial approximation equivalent threat velocity index and real-time position parameters, calculate the spatiotemporal pressure coupling coefficient, the spatiotemporal pressure coupling coefficient is used to characterize the resource competition relationship between the gate's rated total closing time and the time required for the moving target to reach the gate; Spatiotemporal pressure coupling coefficient for: The rated total height of the door. The rated closing speed set for the door. This represents the real-time vertical distance between the moving target and the door plane. To achieve an equivalent safe approximation of the velocity constant, The overall system response delay time. To approximate the equivalent threat velocity index radially, The rated total closing time of the door. The target equivalent safe arrival time threshold; S4 determines the safety intervention probability based on the spatiotemporal pressure coupling coefficient, and outputs the corresponding control signal to the industrial door controller according to the numerical range of the safety intervention probability to perform safety maintenance, early warning deceleration or emergency braking operations.

2. The industrial door anti-pinch safety early warning method based on video stream moving target detection according to claim 1, characterized in that, The direction coefficient is the larger of the cosine of the angle between the world coordinate velocity vector of the moving target and the normal direction of the gate and zero. When the angle is less than 90 degrees, the direction coefficient is positive, indicating that the target is approaching the gate. When the angle is greater than or equal to 90 degrees, the direction coefficient is zero, indicating that the target is moving parallel to or away from the gate.

3. The industrial door anti-pinch safety early warning method based on video stream moving target detection according to claim 1, characterized in that, The equivalent safe approach velocity constant is used to prevent the denominator from being zero when the radial approach equivalent threat velocity exponent is zero, and characterizes the minimum safe buffer threshold for stationary targets.

4. The industrial door anti-pinch safety early warning method based on video stream moving target detection according to claim 1, characterized in that, The formula for calculating the probability of security intervention is: In the formula, To ensure the probability of safe intervention, This is the risk critical threshold constant. Sensitivity index This is the spatiotemporal pressure coupling coefficient.

5. The industrial door anti-pinch safety early warning method based on video stream moving target detection according to claim 4, characterized in that, The sensitivity index is used to adjust the steepness of the response of the probability of security intervention to changes in the spatiotemporal pressure coupling coefficient; The risk critical threshold constant is used to define the critical equilibrium point when the gate closing time is equal to the target impact time.

6. The industrial door anti-pinch safety early warning method based on video stream moving target detection according to claim 1, characterized in that, Preprocessing and feature tracking of real-time video streams, including: Distortion correction is performed on the captured video frames to restore the true geometric proportions of the scene; Gaussian filtering is applied to the corrected video frames to reduce interference from ambient light and sensor noise. The sparse optical flow algorithm is used to track feature corner points in video frames, and the pixel velocity of the moving target in the image coordinate system is obtained by calculating the displacement of the feature corner points between consecutive frames.

7. The industrial door anti-pinch safety early warning method based on video stream moving target detection according to claim 1, characterized in that, Mapping image coordinates to the world coordinate system includes: During the system initialization phase, the camera's intrinsic and extrinsic parameter matrices are determined using a calibration method, and a homography matrix mapping relationship between pixel coordinates and ground world coordinates is constructed. Based on the homography matrix, the pixel displacement of the moving target on the image plane is converted into the instantaneous movement speed on the real world ground in real time; Based on the homography matrix, calculate the vertical distance between the center of the feature point of the moving target and the plane where the door is located.

8. The industrial door anti-pinch safety early warning method based on video stream moving target detection according to claim 1, characterized in that, Output the corresponding control signals to the industrial door controller, including: A first security probability threshold and a second security probability threshold are preset, wherein the second security probability threshold is greater than the first security probability threshold; When the probability of safety intervention is less than the first safety probability threshold, it is determined to be a safe zone, and a signal to maintain the normal closure of the door is output. When the probability of safety intervention is greater than or equal to the first safety probability threshold and less than the second safety probability threshold, it is determined to be a warning interval, and a signal is output to trigger the warning light to flash and control the door to slow down and close. When the probability of safety intervention is greater than or equal to the second safety probability threshold, it is determined to be an emergency stop zone, and a signal to cut off the downward movement of the door and control the door to open in the reverse direction is output.

Images