Rail transit station platform door laser and visual technology protection method

Abstract

The application discloses a kind of rail transit station platform door laser and visual technology protection method, comprising the following steps: semiconductor laser is shaped after collimation, and laser beam is emitted in continuous output mode by beam expander;Laser beam returns scattered light by environmental medium, and scattered light is formed into light spot by band-pass filter to suppress ambient light;Light spot is received by industrial camera, and after preliminary processing of exposure, gain, gamma, data is converted and transmitted into PC end by high-speed link.The laser beam emission technology of the application is based on the straight-line propagation characteristics of laser beam, and the receiving end receives laser beam by emitting laser beam from the emission end, to form light spot.Visual detection method is used simultaneously, camera and computer are used to replace human eye, to monitor the gap between platform door and train door in real time.When foreign matter blocks characteristic light band, the system judges the existence of foreign matter by image processing algorithm, and the device has strong anti-interference ability, good stability and fast response speed.

Description

Technical Field

[0001] This invention relates to the field of rail transit safety facilities technology, specifically a protection method for rail transit station platform doors using laser and vision technology. Background Technology

[0002] In rail transit systems, the safety protection of the gap between platform screen doors and train doors is a crucial aspect of ensuring passenger safety. Traditional foreign object detection devices for rail transit platform screen doors generally use the principle of infrared beam sensors. They emit and receive infrared beams to form a detection grid, and an alarm is triggered when a foreign object blocks the beam. While these devices are low-cost, easy to install, and suitable for simple scenarios, they are prone to false alarms due to ambient light interference (such as light interference or dust) and dust obstruction. They also have poor anti-interference capabilities and low detection accuracy.

[0003] This case arose in order to resolve the aforementioned issues. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a protection method against laser and vision technologies used in rail transit station platform screen doors, solving the problems mentioned in the background section. To achieve the above objectives, this invention is implemented through the following technical solution: A protection method against laser and vision technologies used in rail transit station platform screen doors, comprising the following steps: S1, Working end: A semiconductor laser emits a laser beam in continuous output mode after collimation and beam expansion; the laser beam returns as scattered light through the ambient medium, and the scattered light is filtered by a bandpass filter to suppress ambient light and form a light spot; S2, Receiving end: The light spot is received by an industrial camera, and after preliminary processing of exposure, gain, and gamma, the data is converted and transmitted to a PC via a high-speed link; S3, PC processing: Upon startup, distortion correction and brightness calibration are first performed using a calibration board; the acquisition thread and UI thread are separated, and zero-copy feedback is achieved through a circular buffer; in case of an anomaly, a watchdog timer is triggered to resample and ensure continuity; the aforementioned industrial camera and semiconductor laser are located on opposite sides of the platform screen door (i.e., both sides of the platform screen door).

[0005] The semiconductor laser emits a 940nm laser beam. A filter is installed in front of the industrial camera to filter out all light outside the 940nm ± 10nm range. The industrial camera's camera is an infrared camera that captures light in the 700-1100nm range. The laser beam from the opposite side is directed into the dark cavity to form a spot, preventing ambient light from entering. Furthermore, the industrial camera is tilted within the dark cavity, with the lens tilted downwards at a 45° angle, effectively reducing light interference.

[0006] The aforementioned industrial camera and semiconductor laser are located on opposite sides of the platform door. When a foreign object obstructs the characteristic light band inside the platform door (in case of anomaly), the PC terminal determines the presence of the foreign object through image processing and algorithms.

[0007] As a preferred solution, the PC-side image processing further involves: separating the light band area from the background by setting a brightness threshold, and then using morphological operations to remove noise and fill the light band area; when a foreign object blocks the light band, the image after threshold segmentation will show broken or shadowed areas, and the presence of a foreign object can be determined by detecting these anomalies.

[0008] As a preferred option, the image processing algorithm is further as follows:

[0009] Spot localization (subpixel) (1)

[0010] ;

[0011] (1) Centroid of light spot: where χ and y represent the horizontal and vertical coordinates of a single pixel within the ROI (in pixels), and I(χ,y) is the grayscale or intensity value of the pixel (0-255 for 8-bit, 0-4095 for 12-bit).

[0012] The notation χ represents the summation over all pixels within the ROI; c ,y c These are the horizontal and vertical coordinates of the spot center calculated by the gray-scale centroid method, respectively, which can achieve sub-pixel accuracy.

[0013] Spot size estimation (2)

[0014] ;

[0015] (2) In this equation, r is the equivalent radius of the light spot, used to measure the size of the light spot; χ c ,y c The centroid result is taken from formula (1); the other symbols χ, y, I(χ,y) and the summation symbol have the same meaning as in formula (1);

[0016] Morphological fitting (3)

[0017] ;

[0018] (3) Two-dimensional Gaussian fitting: In the formula, I(χ,y) is the pixel intensity after filtering or normalization; I0 is the Gaussian peak amplitude (brightness of the spot center); (χ0,y0) is the coordinate of the center of the fitted spot; σ is the standard deviation of the Gaussian distribution, used to reflect the spot width; B is the uniform background term, which compensates for ambient light or sensor dark current;

[0019] Occlusion detection (4)

[0020] ;

[0021] (4) Occlusion detection and debouncing: middle, The adaptive threshold for frame t; and These are the average gray level and standard deviation of the ROI in that frame, respectively; k is an empirical coefficient (usually 0.5-1.0). The subsequent exponential moving average formula... middle, α is the smoothing factor (0 < α ≤ 1). The EMA value is the value of the previous frame; if N consecutive frames satisfy S t <T t This means that the laser beam is determined to be blocked. De-jitter conditions. If the interval between two state transitions is less than the minimum jitter time threshold If the time interval is 50 ms, then ignore it;

[0022] Noise suppression (5):

[0023] ;

[0024] (5) Noise suppression: Temporal median filtering is used to remove noise in a 5-frame window; followed by spatial Gaussian convolution. middle, These are the pixels after time-domain filtering. For the kernel shift index, The standard deviation is A discrete Gaussian kernel is used to smooth spatial noise.

[0025] The overall concept of the invention: In rail transit systems, the safety protection of the gap between platform screen doors and train doors is a critical link in ensuring passenger safety. Traditional mechanical or photoelectric sensors have problems such as blind spots and high false alarm rates, while the combination of laser protection and vision technology can significantly improve detection accuracy and reliability. A laser device (transmitter and receiver) is used to shoot light at each other, and a high-definition industrial camera is deployed to collect the light spot.

[0026] Simultaneously, visual algorithms are used to analyze light spot images in real time to identify obstacles.

[0027] By adopting the above technical solution, the protection method for platform screen doors of rail transit stations provided by this invention using laser and vision technology has the following advantages compared with the prior art: Laser beam transmission technology is based on the linear propagation characteristics of laser beams. A laser beam is emitted from the transmitting end and received by the receiving end, forming a light spot. Simultaneously, a vision-based detection method is used, employing a camera and computer to replace the human eye, monitoring the gap between the platform screen door and the train door in real time. When a foreign object obstructs the characteristic light band, the system determines the presence of the foreign object through image processing algorithms. The device has strong anti-interference capabilities, good stability, and fast response speed. Attached Figure Description

[0028] Figure 1Flowchart of a protection method using laser and vision technology for platform doors in rail transit stations. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments are described below, with reference to the appendix. Figure 1 The present invention will be further described in detail below.

[0030] Equipment Installation: Align the transmitter and receiver. Adjust the beam to ensure no obstructions. Flexible Installation: The laser beam's emission angle and distance can be adjusted to suit different scenarios.

[0031] High precision: The linear propagation characteristics of the laser beam determine its high positioning accuracy, making it suitable for scenarios requiring precise detection; Strong anti-interference capability: The laser beam is not easily affected by ambient light or electromagnetic interference, exhibiting good stability; Fast response speed: The laser beam system has a short response time, enabling rapid detection of obstruction events; Strong adaptability: It can adapt to complex environments, such as changes in lighting and background interference; Rich information: It can acquire detailed information such as the shape, color, and texture of the target.

[0032] During the experiment, the brightness of the light spot was compared with the set value (adjustable) to obtain the result. Specifically, two beams of light and two light plates were used simultaneously, each with a set threshold. Reaching the set value indicated that the experiment was successful (in actual operation, the aperture darkens when there is obstruction and brightens when there is no obstruction). It should be noted that the two beams of light emitted at different speeds have a height difference, that is, one high and one low, corresponding to a tall adult and a short child, respectively.

[0033] Working principle: Laser beam transmission technology is based on the linear propagation characteristic of laser beams. A laser beam is emitted from the transmitter and received by the receiver, forming a light spot. Vision-based detection uses cameras and computers to replace human eyes, monitoring the gap between platform screen doors and train doors in real time. When a foreign object obstructs the characteristic light band, the system uses image processing algorithms to determine its presence.

[0034] The visual algorithm (image processing algorithm) of this scheme is as follows: by setting a brightness threshold, the light band region is separated from the background, and then morphological operations are used to remove noise and fill the light band region. When a foreign object blocks the light band, the image after threshold segmentation will show broken or shadowed areas. By detecting these anomalies, the presence of a foreign object can be determined. In addition, deep learning and video analysis can be combined to improve the accuracy of foreign object detection.

[0035] The light spot is emitted in continuous output mode from a semiconductor laser after collimation and beam expansion. The scattered light returning through the working medium is first suppressed by a bandpass filter to suppress ambient light, and then received by an industrial camera (CMOS, global shutter) through a C-mount lens. After the FPGA in the camera performs preliminary processing such as exposure, gain, and gamma, it converts the RAW data into 8 / 12-bit RGB and sends it to the PC via a USB 3.0 high-speed link. The host computer SDK maps the frame buffer to the acquisition software; the software calls the DMA data stream for real-time display, synchronous archiving, and subsequent algorithm modules.

[0036] Upon startup, distortion correction and brightness calibration are first performed using a calibration board; the acquisition thread and UI thread are separated, and zero-copy is achieved through a circular buffer; in case of an anomaly, a watchdog timer is triggered for resampling to ensure continuity. The entire process latency is less than 30ms, which meets the requirements for dynamic monitoring.

[0037] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A protection method of rail transit station platform door laser and vision technology, characterized in that, Comprising the following steps: S1, PC end processing: first use the calibration plate to complete the distortion correction and brightness calibration when starting; acquisition thread Separate from the UI thread, realize zero-copy feedback through the ring buffer; S2, working end: the semiconductor laser is shaped by collimation and beam expansion mirror and emits a laser beam in continuous output mode; the scattered light of the laser beam returned through the environment medium forms a light spot after the ambient light is suppressed by the band-pass filter; S3, receiving end: the light spot is received by the industrial camera, and after preliminary processing of exposure, gain and gamma, the data is converted and transmitted to the PC end through the high-speed link; S4, when the foreign matter appears in the platform door and blocks the characteristic light band--abnormal, the PC end judges the existence of foreign matter through image processing and algorithm, and triggers the watchdog to re-sample to ensure continuity; The above industrial camera and semiconductor laser are located on the opposite side of the platform door; PC end image processing: separate the light band area from the background by setting the brightness threshold, and then remove noise and fill the light band area by morphological operation; When the foreign matter blocks the light band, the threshold segmented image will have a fracture or shadow area, and the existence of foreign matter can be judged by detecting these abnormalities. The laser beam emitted by the semiconductor laser is 940nm, and a filter is installed in front of the industrial camera to filter out light outside 940nm±10nm. The camera of the industrial camera is an infrared camera that collects 700-1100nm light. The lens of the industrial camera is tilted downward and placed in a dark chamber, with an inclination angle of 45°, and the laser beam on the opposite side is shot into the dark chamber to form a light spot.

2. The protection method of rail transit station platform door laser and vision technology according to claim 1, characterized in that:

5. The protection of the rail transit station platform door laser and visual technology according to claim 1 3. The protection method of rail transit station platform door laser and vision technology according to claim 2, characterized in that The image processing algorithm is as follows:

4. The protection method of rail transit station platform door laser and vision technology according to claim 3, characterized in that Light spot positioning (sub-pixel) (1) Light spot size estimation (2) A method, characterized in that Morphological fitting (3) Obstruction judgment (4) (1) Centroid of light spot: where χ and y represent the horizontal and vertical coordinates (in pixel) of a single pixel within the ROI, and I(χ, y) is the gray value (0-255 for 8-bit, 0-4095 for 12-bit) of the pixel; The symbol ∑ denotes summation over all pixels within the ROI; The horizontal and vertical coordinates of the light spot center calculated by the gray centroid method are and, respectively, which can achieve sub-pixel accuracy. Noise suppression (5) (2) where r is the equivalent radius of the spot, which measures the spot size; The centroid result taken from equation (1); the remaining symbols χ, y, I(χ, y) and the summation symbol are the same as in equation (1); ​ (3) Two-dimensional Gaussian fitting: where I(x, y) is the filtered or normalized pixel intensity; is the Gaussian peak amplitude (center of the spot brightness); is the fitted center of the spot coordinates; is the standard deviation of the Gaussian distribution, reflecting the spot width; B is the uniform background term, compensating for ambient light or sensor dark current; ​ (4) Occlusion determination and de-bouncing: In this case, is the adaptive threshold for the t-th frame; and are the average gray level and its standard deviation of the ROI of the frame, respectively; k is an empirical coefficient (usually 0.5-1.0), and the subsequent exponential moving average formula In this case, a is a smoothing factor (0 < a ≤ 1), is the EMA value of the previous frame; If N consecutive frames satisfy That is, it is determined that the laser is blocked, and the dithering condition is met If the interval between two state flips is less than the minimum dithering time threshold Then it is ignored; ​ (5) Noise suppression: temporal median filtering with a 5-frame window to remove noise; followed by spatial Gaussian convolution In some embodiments, is the pixel filtered in the time domain, is the convolution kernel displacement index, is a discrete Gaussian kernel with a standard deviation of used to smooth spatial noise.

Images