Non-dedicated trolley identification and control method and system for escalators / walkways

Abstract

The present invention relates to a non-dedicated trolley identification and control method and system for an escalator. The method comprises: constructing a template image based on a trolley; collecting a front image to be detected; matching the front image with the front template image based on a template matching method, identifying whether there is a trolley in the front image; if no trolley is identified, the escalator operates normally; if the identification result shows that there is a trolley, a side image of the trolley to be detected is collected after a voice prompt; trolley features in the side image are extracted based on an edge detection algorithm; the extracted feature positions are compared with the side template image using a template matching method; if the positions match, it is determined to be a dedicated trolley, and the escalator operates normally; if the positions do not match, it is determined to be a non-dedicated trolley, a control instruction is generated, the escalator is shut down, and an alarm is issued. The present invention has a high recognition rate and improves safety performance.

Description

Technical Field

[0001] The present invention relates to the technical field of elevator applications, and in particular to a method and system for identifying and controlling a non-dedicated trolley for an escalator / walkway. Background Art

[0002] People often push various trolleys from elevators onto escalators or sidewalks, which often causes safety problems. To solve this problem, the existing technology usually adopts the method of setting up safety warning signs near each escalator entrance that read "No use of non-dedicated trolleys";

[0003] Although the signs are obvious, many passengers still bring non-dedicated trolleys onto escalators / walkways, posing a safety hazard. Summary of the Invention

[0004] The purpose of the present invention is to provide a non-dedicated trolley identification and control method and system for escalators / walkways, which has a high recognition rate and improves safety performance.

[0005] In order to solve the above technical problems, the technical solution of the present invention is:

[0006] In a first aspect, a method for identifying and controlling a non-dedicated trolley for an escalator / walkway is provided, comprising:

[0007] Step S100: constructing a template image based on the trolley; the template image includes a front template image constructed based on the trolley and a side template image constructed based on the dedicated trolley;

[0008] Step S200: Capture a front image to be detected; match the front image with a front template image based on a template matching method to identify whether there is a trolley in the front image. If the recognition result is that there is a trolley, a voice prompt is given and step S300 is performed; otherwise, no trolley is recognized;

[0009] Step S300: Capture a side image of the cart to be detected; extract the cart features from the side image based on an edge detection algorithm, and compare the extracted feature positions with the side template image using a template matching method. If the positions match, it is determined to be a dedicated cart, no alarm is issued, and the escalator / walkway power is not cut off, and the escalator / walkway operates normally. If the positions do not match, it is determined to be a non-dedicated cart, and a control instruction is generated to cut off the escalator / walkway operation and issue an alarm.

[0010] Furthermore, in steps S200 and S300, when the front image or the side image is collected, the image is preprocessed; the image preprocessing includes:

[0011] Generate high-resolution images from low-resolution images based on super-resolution methods;

[0012] The image is corrected based on the image deformation method.

[0013] Furthermore, a method for generating a high-resolution image from a low-resolution image based on a super-resolution method includes:

[0014] Calculate feature representation based on the input image and use deep convolutional neural network for feature extraction;

[0015] Combine the feature signals of the input low-resolution image with the weights in the super-resolution model to generate a high-resolution feature map;

[0016] The reconstructed high-resolution feature map is input into the learned inverse transformation network to generate the final high-resolution output image; the formula of the inverse transformation network is as follows:

[0017]

[0018] Among them, X represents the input low-resolution image, W represents the weight matrix, represents the feature extraction function, θ1 and θ2 are the parameters of the two sub-networks respectively, f represents the inverse conversion function, and Y represents the output high-resolution image.

[0019] Furthermore, the method for correcting the image based on the image deformation method specifically includes: affine transformation and perspective transformation; wherein

[0020] Affine transformation is a transformation that keeps straight and parallel lines in the original image unchanged. It is calculated by three anchor points and their positions in the new image.

[0021] A perspective transformation performs rotation, translation, and scaling like an affine transformation, but also preserves the linearity of line segments in the image. It is calculated using four anchor points and their positions in the new image.

[0022] Furthermore, the image preprocessing step also includes:

[0023] Enhance the image;

[0024] Smoothing and denoising images

[0025] Grayscale the image;

[0026] Perform threshold segmentation and binarization on the image.

[0027] Furthermore, in steps S200 and S300 , the template matching method adopts a template matching algorithm based on correlation measurement between pixels in OpenCV.

[0028] Furthermore, in step S300, the method of extracting the cart features from the side image using the edge detection algorithm is as follows:

[0029] Perform edge detection on the input image;

[0030] Perform contour extraction on the obtained edge;

[0031] Shape matching is performed on each contour to determine the location of each cart feature.

[0032] Furthermore, the obtained edges are subjected to contour extraction by using the findContours function in OpenCV to extract the contours of the edge image. The specific steps are as follows:

[0033] (1) Ensure that the input image is a binary image. If not, threshold segmentation and binarization operations need to be performed first;

[0034] (2) Use the Canny operator edge detection algorithm to obtain the edge image of the target object;

[0035] (3) Before calling the findContours function, you need to create an empty image as the contour output result and a blank matrix for storing hierarchical information;

[0036] (4) Call the findContours function to get the contour;

[0037] (5) Traverse each contour, calculate the feature quantity, and prepare for further processing.

[0038] Furthermore, the method for shape matching of each contour is:

[0039] (1) Determine the outline of the target object;

[0040] (2) Applying a rectangular fitting function or a circle fitting function to each contour, the regional moments of the contour, i.e., the first-order and second-order moments of the image, are calculated to obtain the center coordinates and basic area characteristics of the contour;

[0041] (3) Obtaining the fitting rectangle or fitting circle center information by calculating the contour feature quantity;

[0042] (4) Determine the position, size and shape information of the target object based on the fitting results.

[0043] In a second aspect, a non-dedicated trolley identification and control system for an escalator / walkway is provided, comprising:

[0044] The forward-facing camera is located in front of the escalator entrance and is used to capture images in front of the escalator entrance.

[0045] Side cameras are located on the side of the escalator entrance to capture images of the trolley's side;

[0046] The controller is connected to the front camera and the side camera and is used to obtain the collected images; it is used to match the front image with the front template image based on the template matching method, and identify whether there is a trolley in the front image. If the trolley is not identified, the escalator operates normally. If the recognition result is that there is a trolley, after a voice prompt, the trolley features of the side image are extracted based on the edge detection algorithm, and the extracted feature positions are compared with the side template image through the template matching method. If the positions match, it is determined to be a dedicated trolley, no alarm is given, and the escalator / walkway power is not cut off, and the escalator / walkway operates normally; if the positions do not match, it is determined to be a non-dedicated trolley, and a control instruction is generated to cut off the operation of the escalator / walkway and issue an alarm.

[0047] The alarm is connected to the controller for voice prompts or alarms.

[0048] The present invention has the following beneficial effects:

[0049] First, the present invention records a front image and a side image at the entrance of an escalator / moving walkway. The front image is matched with a front template image based on a template matching method to identify whether there is a trolley in the front image. If no trolley is identified, the escalator operates normally. If the identification result is a trolley, a voice prompt is given, and the trolley features in the side image are extracted based on an edge detection algorithm. The extracted feature positions are compared with the side template image using a template matching method. If the positions match, it is determined to be a dedicated trolley, no alarm is issued, and the escalator / moving walkway power is not cut off, and the escalator / moving walkway operates normally. If the positions do not match, it is determined to be a non-dedicated trolley, and a control command is generated to cut off the escalator / moving walkway operation and issue an alarm. The present invention has the characteristics of being able to identify non-dedicated trolley features and issue a real-time alarm and control the escalator to stop, thereby preventing non-dedicated trolleys from illegally riding on escalators / moving walkways and preventing safety accidents.

[0050] 2. The image preprocessing method of the present invention significantly increases image details and fidelity, corrects deformed images, improves detection accuracy, and improves image recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a flow chart of the identification control method of the present invention;

[0052] Figure 2 This is a schematic diagram of the structure of the identification control system of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0054] Please refer to Figure 1 and Figure 2 ,The present invention is a non-dedicated trolley identification and control system for an escalator / moving walkway, which comprises: a forward-facing camera, arranged in front of the escalator entrance for capturing an image in front of the escalator entrance;

[0055] Side cameras are located on the side of the escalator entrance to capture images of the trolley's side;

[0056] The controller is connected to the front camera and the side camera and is used to obtain the collected images; it is used to match the front image with the front template image based on the template matching method, and identify whether there is a trolley in the front image. If the trolley is not identified, the escalator operates normally. If the recognition result is that there is a trolley, after a voice prompt, the trolley features of the side image are extracted based on the edge detection algorithm, and the extracted feature positions are compared with the side template image through the template matching method. If the positions match, it is determined to be a dedicated trolley, no alarm is given, and the escalator / walkway power is not cut off, and the escalator / walkway operates normally; if the positions do not match, it is determined to be a non-dedicated trolley, and a control instruction is generated to cut off the operation of the escalator / walkway and issue an alarm.

[0057] The alarm is connected to the controller for voice prompts or alarms.

[0058] In this embodiment, the forward-facing and side-facing cameras are from the Daheng Industrial Camera MER-500-7UM / UC series, a USB 2.0 digital camera suitable for a variety of typical machine vision applications, such as surface inspection, positioning, defect detection, and measurement. They utilize a 1 / 2.5" line-exposure CMOS sensor chip and are extremely compact, measuring only 29mm x 29mm x 29mm, making them suitable for installation at escalator and sidewalk entrances without obstructing passenger access. They also feature an integrated I / O interface and a cable locking device, ensuring stable operation in harsh environments. These highly reliable, cost-effective industrial digital cameras transmit captured images via USB 2.0 to a remote controller, which then begins processing the images.

[0059] The controller uses the Intel Celeron J6413 / x6425E processor from the AVS-32X series industrial computers, enabling object positioning, identification, guidance, measurement, and inspection. The processor can simultaneously receive image signals from different escalators in the same area (shopping malls, buildings), analyze and identify the corresponding images, and then provide feedback based on the recognition results to control escalator-side voice-activated warnings or elevator control.

[0060] The Yuexin YX75R alarm uses a USB-to-TTL or 485 control interface. It features a compact size and lightweight design, making it ideal for escalator entrances. It features a high-brightness LED flash and a loud, customizable alarm tone. It offers long-term continuous operation, reliable and stable operation, and a variety of mounting options, including screw mounting and magnetic attachment.

[0061] The present invention also provides a method for identifying and controlling non-dedicated trolleys for escalators / walkways, which uses the above-mentioned method for identifying and controlling non-dedicated trolleys for escalators / walkways, and the method comprises:

[0062] Step S100: constructing a template image based on the trolley; the template image includes a front template image constructed based on the trolley and a side template image constructed based on the dedicated trolley;

[0063] Step S200: Capture a front image to be detected; match the front image with a front template image based on a template matching method to identify whether there is a trolley in the front image. If the recognition result is that there is a trolley, a voice prompt is given and step S300 is performed; otherwise, no trolley is recognized;

[0064] Step S300: Capture a side image of the cart to be detected; extract the cart features from the side image based on an edge detection algorithm, and compare the extracted feature positions with the side template image using a template matching method. If the positions match, it is determined to be a dedicated cart, no alarm is issued, and the escalator / walkway power is not cut off, and the escalator / walkway operates normally. If the positions do not match, it is determined to be a non-dedicated cart, and a control instruction is generated to cut off the escalator / walkway operation and issue an alarm.

[0065] The following are Figure 1 Each step in is described in detail.

[0066] In step S100, a trolley is placed 3 meters away from the escalator entrance, and the front-facing camera at the escalator / walkway entrance collects images and processes them through the controller to construct a front template image;

[0067] A special trolley is placed at the entrance of the escalator. The side camera collects images and uses the controller to identify the features of each component of the special trolley (handle, frame / seat / recliner, tire), and records the coordinate position to form a side template image.

[0068] In step S200, the forward camera at the escalator / sidewalk entrance captures and recognizes the front image 3 meters away from the escalator entrance at preset intervals; in step S300, the side camera recognizes the side image of the trolley.

[0069] In steps S200 and S300 , when a front image or a side image is collected, the image is pre-processed.

[0070] Image preprocessing includes: generating high-resolution images from low-resolution images based on super-resolution methods and correcting images based on image deformation methods.

[0071] Among them, super-resolution methods are used to generate high-resolution images from low-resolution images. Deep learning technology is used to generate high-resolution images from low-resolution images. It can significantly increase image details and fidelity, thereby improving the accuracy of computer vision, medical imaging and other fields. Super-resolution technology is a preprocessing technology that reconstructs high-resolution output images from low-resolution input images through deep learning or non-learning methods. Specifically:

[0072] (1) Feature extraction: Calculate feature representation based on the input image and use deep convolutional neural network for feature extraction;

[0073] (2) Reconstructing feature maps: combining the feature signals of the input low-resolution image with the weights in the super-resolution model to generate a high-resolution feature map;

[0074] (3) Reconstructing a high-resolution image: The reconstructed high-resolution feature map is input into the learned inverse transformation network to generate the final high-resolution output image. The formula of the inverse transformation network is as follows:

[0075]

[0076] Among them, X represents the input low-resolution image, W represents the weight matrix, represents the feature extraction function, θ1 and θ2 are the parameters of the two sub-networks respectively, f represents the inverse conversion function, and Y represents the output high-resolution image.

[0077] Among them, correcting an image based on an image deformation method refers to performing a series of spatial transformations (including rotation, translation, scaling, distortion, etc.) on the original image to obtain a new deformed image, specifically including: affine transformation and perspective transformation; among them, affine transformation is a transformation that keeps the straight lines and parallel lines in the original image unchanged, and is calculated by three positioning points and their positions in the new image; perspective transformation performs rotation, translation and scaling like affine transformation, but can also keep the line segments in the image linear, and is calculated by four positioning points and their positions in the new image.

[0078] Image preprocessing also includes:

[0079] Enhance the image;

[0080] Smoothing and denoising images

[0081] Grayscale the image;

[0082] Perform threshold segmentation and binarization on the image.

[0083] In image enhancement, the emphasize operator is used to enhance image details, making the edges of the cart in the photo clearer and the features easier to identify. The principle is as follows:

[0084] (1) Use low-pass (mean_image) for filtering, the mask size is askWidth*Mask Height, the larger the matrix, the lower the crossover frequency;

[0085] (2) Obtain the grayscale value (mean) of the filtered image;

[0086] (3) Based on the original image grayscale value (orig) and (mean), the new image grayscale value (new) is calculated as follows: new = round ((orig - mean) * Factor) + orig; where Factor is the coefficient of contrast increase;

[0087] Grayscale values ​​that overflow and / or underflow are clipped.

[0088] Among them, in the image smoothing and denoising, the median filter is used to filter and denoise the collected pictures, and the filtering method in the spatial domain is adopted. All the pixels in the neighborhood of the selected cart pixel are processed through the algorithm, and the processing result is used as the output value of this pixel.

[0089] Among them, when graying the image, the operator rgb1 to gray (Image Median, GrayImage) can be used to gray the cart image. Considering that if the light intensity is weak, the gray value of the collected image is mainly distributed in the low gray value area, which will directly affect the quality of image processing. The gamma correction method is used to perform nonlinear correction on the grayscale of the image. The basic form is as follows:

[0090] g=cq γ

[0091] Where g is the pixel value of the output image, q is the pixel value of the input image, c is the proportional coefficient, which is generally 1, and γ is the power exponent of the Gamma correction. When γ < 1, the grayscale is expanded, and when γ > 1, the grayscale is compressed.

[0092] Among them, when performing threshold segmentation and binarization on the image, the cart image can be threshold segmented after being grayscaled, and the histogram valley bottom determination threshold method is used to separate the cart image from the background, so as to achieve the purpose of extracting the target image.

[0093] After segmenting the cart from the background, the dyn-threshold operator is applied. This operator is suitable for situations where a single grayscale image cannot be used for segmentation. For example, the background grayscale is complex, with some parts brighter and some darker than the foreground object. Alternatively, the foreground object may contain multiple grayscales, making it difficult to segment using a global threshold. This operator is suitable for the complex background of the cart image. The cart can be successfully segmented from the background.

[0094] In steps S200 and S300, the method for extracting features of the cart from the side image based on the edge detection algorithm includes:

[0095] 1. Perform edge detection on the input image; for example, use edge detection algorithms such as the Canny operator;

[0096] 2. Perform contour extraction on the obtained edge; for example, use the findContours function in OpenCV; use the findContours function in OpenCV to extract the contour of the edge image. The specific steps are as follows:

[0097] (1) Ensure that the input image is a binary image. If not, threshold segmentation and binarization operations need to be performed first;

[0098] (2) Use the Canny operator edge detection algorithm to obtain the edge image of the target object;

[0099] (3) Before calling the findContours function, you need to create an empty image as the contour output result and a blank matrix for storing hierarchical information;

[0100] (4) Call the findContours function to get the contour; the syntax is as follows:

[0101] contours,hierarchy=cv.findContours(image,mode,method[,contours[,hierarchy[,offset]]])where image is the input binary image; mode determines the contour retrieval mode (usually RETR_EXTERNAL is selected to obtain only the outermost contour); method determines the contour approximation method (usually CHAIN_APPROX_SIMPLE is selected to retain only the endpoints of the contour). At the same time, contours is used to save the detected contours, and hierarchy is used to save the hierarchical relationship information between contours.

[0102] (5) Traverse each contour, calculate its area, perimeter, center of gravity and other feature quantities, and prepare for further processing.

[0103] 3. Perform shape matching on each contour to determine the location of each cart feature. When further processing the obtained contour, a shape matching algorithm can be used to determine the location, size, and shape of the target object. Commonly used shape matching algorithms include rectangle fitting algorithm and circle fitting algorithm. The specific steps are as follows:

[0104] (1) Determine the outline of the target object;

[0105] (2) For each contour, the rectangular fitting function or the circle fitting function cv::moments is used to calculate the regional moments of the contour, that is, the first-order and second-order moments of the image, so as to obtain the basic feature quantities such as the center coordinates and area of ​​the contour;

[0106] (3) Obtain the fitted rectangle or fitted circle center information by calculating the contour feature quantity; OpenCV provides the following functions:

[0107] Rectangle fitting function: cv::minAreaRect();

[0108] Circle fitting function: cv::minEnclosingCircle();

[0109] (4) Determine the position, size and shape information of the target object based on the fitting results.

[0110] In steps S200 and S300 , the template matching method adopts a template matching algorithm based on correlation measurement between pixels in OpenCV.

[0111] The front image is matched against a front template image using a template matching method to identify whether a trolley is present. If no trolley is detected, the escalator operates normally. If a trolley is detected, an audible warning is issued, stating, "No non-dedicated trolleys are allowed on escalators / walkways." The system then waits for the side camera to capture a side image. The edge detection algorithm extracts the trolley's features from the side image. Template matching is used to compare the extracted features with the side template image. If the positions match, the trolley is identified as a dedicated one, and no alarm or power is cut off for the escalator / walkway, allowing it to operate normally. If the positions do not match, the trolley is identified as a non-dedicated one, and a control command is generated to cut off the associated escalator / walkway safety circuit, causing the escalator / walkway to stop and an alarm to be issued.

[0112] The present invention aims to overcome the existing technology and the urgent need for a safe and reliable escalator / sidewalk trolley safety warning and chain reaction method system on the market. It provides an intelligent identification and escalator control method and system that can use machine vision to identify the characteristics of trolleys and prevent non-dedicated trolleys from illegally riding on escalators / sidewalks, thereby preventing non-dedicated trolleys from illegally riding on escalators / sidewalks and preventing safety accidents.

[0113] The parts not involved in the present invention are the same as the existing technology or are implemented by using the existing technology.

[0114] The above content is a further detailed description of the present invention in conjunction with specific embodiments, and the specific implementation of the present invention should not be considered to be limited to these descriptions. For those skilled in the art of the present invention, without departing from the concept of the present invention, several simple deductions or substitutions can be made, which should be considered to fall within the scope of protection of the present invention.

Claims

1. A method for identifying and controlling non-dedicated trolleys for escalators and sidewalks, characterized by: include Step S100: constructing a template image based on the trolley; the template image includes a front template image constructed based on the trolley and a side template image constructed based on the dedicated trolley; Step S200: Capture a front image to be detected; match the front image with a front template image based on a template matching method to identify whether there is a trolley in the front image. If the recognition result is that there is a trolley, a voice prompt is given and step S300 is performed; otherwise, no trolley is recognized; Step S300: Capture a side image of the cart to be detected; extract cart features from the side image using an edge detection algorithm, and compare the extracted feature positions with the side template image using a template matching method. If the positions match, the cart is determined to be a dedicated cart, and no alarm is issued or the escalator / walkway power is disconnected, allowing the escalator / walkway to operate normally. If the positions do not match, the cart is determined to be a non-dedicated cart, and a control command is generated to disconnect the escalator / walkway and issue an alarm. In step S300, the method of extracting the features of the trolley from the side image using the edge detection algorithm is as follows: Perform edge detection on the input image; Perform contour extraction on the obtained edge; Shape matching is performed on each contour to determine the location of each cart feature.

2. The non-dedicated trolley identification and control method for escalators / walkways according to claim 1, characterized in that: In steps S200 and S300, when a front image or a side image is acquired, the image is preprocessed. The image preprocessing includes: Generate high-resolution images from low-resolution images based on super-resolution methods; The image is corrected based on the image deformation method.

3. The non-dedicated trolley identification and control method for escalators / walkways according to claim 2, characterized in that: Methods for generating high-resolution images from low-resolution images based on super-resolution methods include: Calculate feature representation based on the input image and use deep convolutional neural network for feature extraction; Combine the feature signals of the input low-resolution image with the weights in the super-resolution model to generate a high-resolution feature map; The reconstructed high-resolution feature map is input into the learned inverse transformation network to generate the final high-resolution output image; the formula of the inverse transformation network is as follows: Among them, X represents the input low-resolution image, W represents the weight matrix, represents the feature extraction function, θ1 and θ2 are the parameters of the two sub-networks respectively, f represents the inverse conversion function, and Y represents the output high-resolution image.

4. The non-dedicated trolley identification and control method for escalators / walkways according to claim 2, characterized in that: The method for correcting an image based on the image deformation method specifically includes: affine transformation and perspective transformation; Affine transformation is a transformation that keeps straight and parallel lines in the original image unchanged. It is calculated by three anchor points and their positions in the new image. A perspective transformation performs rotation, translation, and scaling like an affine transformation, but also preserves the linearity of line segments in the image. It is calculated using four anchor points and their positions in the new image.

5. The non-dedicated trolley identification and control method for escalators / walkways according to claim 2, characterized in that: Image preprocessing steps also include: Enhance the image; Smoothing and denoising images Grayscale the image; Perform threshold segmentation and binarization on the image.

6. The non-dedicated trolley identification and control method for escalators / walkways according to claim 1, characterized in that: In steps S200 and S300 , the template matching method adopts a template matching algorithm based on correlation measurement between pixels in OpenCV.

7. The non-dedicated trolley identification and control method for escalators / walkways according to claim 1, characterized in that: The obtained edge is extracted by using the findContours function in OpenCV to extract the contour of the edge image. The specific steps are as follows: (1) Ensure that the input image is a binary image. If not, threshold segmentation and binarization operations need to be performed first; (2) Use the Canny operator edge detection algorithm to obtain the edge image of the target object; (3) Before calling the findContours function, you need to create an empty image as the contour output result and a blank matrix for storing hierarchical information; (4) Call the findContours function to get the contour; (5) Traverse each contour, calculate the feature quantity, and prepare for further processing.

8. The non-dedicated trolley identification and control method for escalators / walkways according to claim 1, characterized in that: The method for shape matching for each contour is: (1) Determine the outline of the target object; (2) Applying a rectangular fitting function or a circle fitting function to each contour, the regional moments of the contour, i.e., the first-order and second-order moments of the image, are calculated to obtain the center coordinates and basic area characteristics of the contour; (3) Obtaining the fitting rectangle or fitting circle center information by calculating the contour feature quantity; (4) Determine the position, size and shape information of the target object based on the fitting results.

9. Non-dedicated trolley identification and control system for escalators / walkways, characterized by: include: The forward-facing camera is located in front of the escalator entrance and is used to capture images in front of the escalator entrance. Side cameras are located on the side of the escalator entrance to capture images of the trolley's side; The controller is connected to the front camera and the side camera and is used to acquire the collected images; it is used to match the front image with the front template image based on the template matching method, and identify whether there is a trolley in the front image. If the trolley is not identified, the escalator operates normally. If the recognition result is that there is a trolley, after a voice prompt, the trolley features in the side image are extracted based on the edge detection algorithm, and the extracted feature positions are compared with the side template image through the template matching method. If the positions match, it is determined to be a dedicated trolley, no alarm is issued, and the escalator / walkway power is not cut off, and the escalator / walkway operates normally; if the positions do not match, it is determined to be a non-dedicated trolley, and a control instruction is generated to cut off the escalator / walkway operation and issue an alarm; The alarm is connected to the controller for voice prompts or alarms.

Images