An elevator car door blocking behavior detection method and a computer-readable storage medium

Abstract

The present invention relates to a method for detecting the behavior of blocking elevator doors, which includes: S1, obtaining a plurality of video segments to be detected; S2, acquiring the pixel mosaic map of each video segment to be detected and sending it as an input to a preset classification model, and according to the classification result output by the classification model, if the output indicates that there may be a door-blocking behavior, then proceed to step S3, if the output indicates that there is no door-blocking behavior, then proceed to step S4; S3, calculating the structural similarity index between two adjacent frames in the video segment to be detected, and judging the overall structural similarity of the video segment to be detected. If the structural similarity is low, it is judged that there is no door-blocking behavior and proceed to step S4. If the structural similarity is high, it is judged that there is a door-blocking behavior; S4: Return to step S2 and continue to detect whether there is a door-blocking behavior in the next video segment to be detected. It can identify the behavior of blocking elevator doors in elevator monitoring videos and accurately identify the behavior of blocking elevator doors without manual operation.

Description

Technical Field

[0001] The present invention relates to the fields of behavior recognition and intelligent elevators, and particularly to a method for detecting the behavior of blocking the elevator door and a computer-readable storage medium. Background Art

[0002] Car elevators are applied in various high-rise residential communities, office buildings, shopping malls and other places, bringing great convenience to our lives. However, safety accidents caused by some bad behaviors in the elevator cannot be ignored. Among them, the behavior of blocking the door seriously endangers the safe operation of the elevator. Some passengers use their bodies or other objects to block the elevator door, causing great potential safety hazards.

[0003] In order to enable elevator management personnel to detect the occurrence of the bad behavior of blocking the door in the elevator in a timely manner, there are already some studies on the elevator door switch failure. One type of method is to use a series of line detection methods such as the Hough line detection principle to perform line detection on the elevator door edge, perform edge processing and binarization operations on the image, and then judge the elevator door failure state through the distance change between the elevator door edge lines. This detection method is simple and has a fast calculation speed, but it is greatly affected by the environment and is prone to detecting other lines in the background, resulting in misjudgment; another type of method is to use the pixel value difference between images to judge the door failure state, compare the pixel values of each frame of image with the pixel values in the fully closed state of the elevator door, thereby judging the opening and closing of the door and giving failure information. This method also has the disadvantage of being greatly affected by the ambient light. In addition, the current research on elevator door switch failure only focuses on the opening and closing state of the elevator door itself, and does not pay attention to the behavior feature of blocking the door, and it is easy to identify the situation where a large number of people continuously enter and exit the elevator as a failure state, which is not convenient for the safe operation and management and maintenance of the elevator. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for detecting the behavior of blocking the elevator door and a computer-readable storage medium, which use the established classification model to judge the pixel mosaic map of the video segment to be detected, and further accurately judge whether the video segment to be detected contains the behavior of blocking the door in combination with the overall structural similarity of each video segment to be detected, and identify the behavior of blocking the elevator door in the elevator monitoring video, so as to achieve the effect of accurately identifying the behavior of blocking the elevator door without manual operation.

[0005] To solve the above technical problems, the present invention provides a method for detecting the behavior of blocking the elevator door, including the following steps: S1. Divide the elevator monitoring video stream to be detected into equal parts to obtain multiple video segments to be detected; S2. Obtain the pixel mosaic map of each video segment to be detected, and send it as an input to a preset classification model. According to the classification result output by the classification model, if the output indicates that there may be a door-blocking behavior, go to step S3; if the output indicates no door-blocking behavior, go to step S4; S3. Calculate the structural similarity index between two adjacent frames of the video segment to be detected. According to the structural similarity index, judge the overall structural similarity of the video segment to be detected. If the structural similarity is low, judge that there is no door-blocking behavior and go to step S4; if the structural similarity is high, judge that there is a door-blocking behavior and output the video segment to be detected; S4: Return to step S2 and continue to detect whether there is a door-blocking behavior in the next video segment to be detected until all video segments to be detected are detected.

[0006] Preferably, in S1, "dividing the elevator monitoring video stream to be detected into equal parts" specifically includes: obtaining a sliding window with a length of s, setting the sliding step of the sliding window to q; passing the elevator monitoring video stream to be detected through the sliding window, that is, intercepting a video segment with a length of s every q seconds from the elevator monitoring video stream to be detected.

[0007] Preferably, in S3, the calculation formula for the structural similarity index between two adjacent frames of the video is:

[0008] where u X and u Y respectively represent the average gray values of two adjacent frames of images X and Y, σ X and σ Y respectively represent the standard deviations of the gray values of two adjacent frames of images X and Y, σ X 2 and σ Y 2 respectively represent the variances of the gray values of two adjacent frames of images X and Y, σ XY represents the covariance of the gray values of two adjacent frames of images X and Y, C1 and C2 are constants, taking C1=(K1*L) 2 , C2=(K2*L) 2 , K1 = 0.01, K2 = 0.03, L = 255.

[0009] Preferably, the method for judging the overall structural similarity of the video segment to be detected specifically includes:

[0010] If the structural similarity index of two adjacent frames of a video is detected to be greater than a threshold value, start counting from these two adjacent frames. If the structural similarity indices of L consecutive frames are greater than the threshold value, it is determined that the overall similarity of the video segment to be detected is relatively high; otherwise, it is determined that the overall similarity of the video segment to be detected is relatively low.

[0011] Preferably, in S2, the method for obtaining the pixel mosaic map of each video segment to be detected specifically includes the following steps: Extract m frames of pictures from each video segment to be detected, and extract the nth row of pixels located in the upper half of each frame of picture; Set each nth row of pixels as a one-dimensional vector, and m one-dimensional vectors are extracted from each video segment to be detected; Arrange and splice the m one-dimensional vectors extracted from each video segment to be detected in the column direction in chronological order to form a pixel mosaic map with a width of m.

[0012] Preferably, the selection of the nth row of pixels follows the following steps: Start from the first row of pixels at the uppermost end of each frame of picture drawn, traverse each row of pixels row by row from top to bottom, and calculate the gray-scale mean value of each row of pixels; Obtain the gray-scale mean value of the background pixel row of the picture, and compare the gray-scale mean value of each row of pixels with the gray-scale mean value of the background pixel row in turn; Set a threshold value. When the first pixel row a whose difference from the gray-scale mean value of the background pixel row in the picture is greater than the threshold value is searched, it is determined that the elevator door area starts from the a+1th row of pixel positions, and the a+1th row of pixels is selected as the nth row of pixels.

[0013] Preferably, the classification model is set up as follows: Collect and screen the monitoring videos of the elevator to obtain multiple video segments with door-blocking behaviors and non-door-blocking behaviors; Frame extraction is performed on all video segments one by one, one row of pixels at the same position in each frame is extracted, and the multiple pixel rows extracted are spliced in chronological order to obtain the pixel mosaic map of each video segment; The multiple pixel mosaic maps are divided into two categories: door-blocking behavior and non-door-blocking behavior, the image features in the two categories of pixel mosaic maps are extracted, and the image features are trained to establish a classification model.

[0014] Preferably, the two types of pixel mosaic maps are sent to the MobileNetV3 convolutional neural network for image feature extraction.

[0015] Preferably, the establishment of the classification model further includes: Dividing the two types of pixel mosaic maps of door-blocking behavior and non-door-blocking behavior into a training set and a validation set according to a preset ratio; Inputting the training set into the MobileNetV3 convolutional neural network, using the MobileNetV3 convolutional neural network to extract the image features of the training set, and training the image features to establish a classification model; Using the validation set to verify the accuracy of the classification model.

[0016] A computer-readable storage medium stores a computer program thereon. Preferably, program code is stored in the computer program, and the program code is loaded and executed by a processor to implement the elevator door blocking behavior detection method described above.

[0017] The above technical solution of the present invention has the following advantages compared with the prior art:

[0018] 1. The present invention identifies the door blocking behavior in the elevator based on pixel features. By obtaining the pixel mosaic map of the video segment to be detected, using the established classification model to judge the pixel mosaic map, and further discriminating whether the video segment to be detected contains the door blocking behavior based on the overall structural similarity of each video segment to be detected, the automatic detection of the door blocking behavior in the elevator is realized, and the accuracy of identifying the door blocking behavior in the elevator is improved.

[0019] 2. By outputting the video segments determined to contain the door blocking behavior, the present invention facilitates subsequent further viewing and processing, further reduces the false positive rate, and ensures the accuracy of detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to make the content of the present invention easier to be clearly understood, the following further details the present invention according to specific embodiments of the present invention in conjunction with the drawings, wherein:

[0021] Figure 1 is a schematic diagram of the working process of the present invention;

[0022] Figure 2 is a schematic diagram of pixel row interception in the specific embodiment of the present invention;

[0023] Figure 3 is a schematic diagram of pixel row mosaic in the specific embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The following further describes the present invention in conjunction with the drawings and specific embodiments, so that those skilled in the art can better understand the present invention and be able to implement it, but the embodiments cited do not limit the present invention.

[0025] Referring to Figures 1 to 3 as shown, the present invention discloses an elevator door blocking behavior detection method and a computer-readable storage medium. The above elevator door blocking behavior detection method includes the following steps:

[0026] Establish a classification model: Collect the monitoring videos of the car elevators in high-rise residential buildings and office buildings, and screen and classify the collected multiple video segments into two categories: those containing the door blocking behavior and those without the door blocking behavior;

[0027] Among them, the length of the collected video segments is preferably 10s.

[0028] For all the collected video segments, frame extraction is performed frame by frame respectively, and the frame images obtained by frame extraction for each video segment are arranged in chronological order. Further, for each of the arranged frame images obtained by frame extraction in each video segment, extraction is carried out, extracting a row of pixels at the same position in each frame, and the multiple pixel rows obtained by extraction are spliced in chronological order to obtain a pixel splicing map for each video segment.

[0029] Preferably, m frame images are extracted for each video segment.

[0030] Among them, obtaining the pixel splicing map for each video segment specifically includes the following process: (1) For the m images obtained by frame extraction for each video segment in the above steps, use a python program to extract the nth row of pixels of each image. Each row of pixels extracted is a one-dimensional vector, and a total of m one-dimensional vectors are extracted for each video segment. (2) Arrange and splice the m one-dimensional vectors extracted in each video segment in the column direction in chronological order to form a pixel splicing map with a width of m.

[0031] Further, in the above process (1), use a python program to extract the nth row of pixels of each image. The determination method of the n value is as follows: Starting from the first row of pixels at the uppermost end of the extracted frame images, traverse each pixel row row by row from top to bottom, and compare the gray scale mean value of each row of pixels with the gray scale mean value of the background pixel row of the image in turn. Set a threshold L. Search for the first pixel row a whose difference from the gray scale mean value of the background pixel row is greater than the threshold. Then determine that starting from the a-th row pixel position is the elevator door area, and select the (a + 1)-th row pixel as the nth row pixel.

[0032] The obtained pixel splicing maps of the above multiple video segments are screened and divided into two categories: with door-blocking behavior and without door-blocking behavior according to their source videos. Send the two categories of the above pixel splicing maps into the MobileNetV3 convolutional neural network to extract image features, train the image features, and establish a trained classification model.

[0033] Specifically, the process of establishing the classification model is as follows: (1) The pixel mosaic maps of multiple video segments obtained are divided into two categories according to their source video segments. If the pixel mosaic map is from a video segment containing a door-blocking behavior, the above pixel mosaic map is classified into the "door-blocking" category; if the pixel mosaic map is from a video segment without a door-blocking behavior, the above pixel mosaic map is classified into the "no door-blocking" category. (2) The two categories of pixel mosaic maps are divided into a training set and a validation set according to a preset ratio of 4:1. The pixel mosaic maps of the training set are input into the above MobileNetV3 convolutional neural network. The output layer of the MobileNetV3 convolutional neural network is changed to 2 categories, and the relevant image features of the extracted pixel mosaic maps are used to train the classification model. The accuracy of the classification model is verified using the validation set. The above classification model obtained through training can initially judge the door-blocking behavior in the elevator.

[0034] In another preferred embodiment, preferably, there are 264 pixel mosaic maps in the training set and 66 pixel mosaic maps in the validation set.

[0035] Specific detection of the door-blocking behavior in the elevator:

[0036] Step 1: First, obtain the elevator monitoring video stream A to be detected, and equally divide the above elevator monitoring video stream A to be detected to obtain multiple video segments to be detected. By equally dividing the elevator monitoring video stream A to be detected, it is convenient for subsequent unified processing.

[0037] Among them, a sliding window with a fixed length of s is set, and the sliding step of the above sliding window is set to q. The elevator monitoring video stream to be detected passes through the sliding window, and a video segment to be detected with a fixed length of s is intercepted from the elevator monitoring video stream to be detected every q seconds.

[0038] In another preferred embodiment, the elevator video stream A to be detected passes through a sliding window with a fixed length of 10s and a step of 2s, and the elevator video stream A to be detected is intercepted into multiple video segments with a length of 2s.

[0039] Step 2: Refer to the method in establishing the classification model to obtain the pixel mosaic map of each video segment to be detected, and use the obtained pixel mosaic map as the input and send it to the pre-established classification model. The output classification result is that there may be a door-blocking behavior or no door-blocking behavior. If the output classification result is that there may be a door-blocking behavior, go to Step 3; if the output classification result is no door-blocking behavior, go to Step 4.

[0040] Refer to Figure 2As shown, in another preferred embodiment, it is preferably to extract 250 frames of pictures from each video segment, and use a Python program to extract the 27th row of pixels of each picture. The 250 one-dimensional vectors extracted are arranged and spliced in chronological order along the column direction, referring to Figure 3 as shown, to form a pixel splicing map with a width of 250 pixels.

[0041] Step 3: Implement an auxiliary discrimination method for the multiple video segments to be detected obtained in the above steps. Specifically, calculate the structural similarity index (i.e., SSIM index) of adjacent two frames of monitoring pictures in each video segment to be detected, output the calculation result, and judge the SSIM similarity between adjacent two frames of monitoring pictures in each video segment to be detected.

[0042] Among them, the formula for calculating the SSIM index between adjacent two frames of pictures is:

[0043] Among them, u X , u Y respectively represent the average gray values of images X and Y, σ X , σ Y respectively represent the standard deviations of the gray values of images X and Y, σ X 2 , σ Y 2 respectively represent the variances of the gray values of images X and Y, σ XY represents the covariance of the gray values of images X and Y, C1 and C2 are constants, usually taking C1 = (K1 * L) 2 , C2 = (K2 * L) 2 , taking K1 = 0.01, K2 = 0.03, L = 255, and after calculation, C1 = 6.5025, C2 = 58.5225.

[0044] If the SSIM index calculated between a certain adjacent two frames of pictures is greater than the threshold, and the threshold is preferably 0.95, it is determined that the SSIM similarity between the adjacent two frames of pictures in the video segment to be detected is relatively high; otherwise, it is determined that the SSIM similarity between the adjacent two frames of pictures in the video segment to be detected is relatively low. If it is detected that the SSIM similarity between adjacent two frames of pictures is relatively high, start counting from these adjacent two frames of pictures. If the SSIM similarity of L consecutive frames of adjacent pictures is relatively high, and L is preferably 150, it is determined that the overall SSIM similarity of the video segment to be detected is relatively high; otherwise, it is determined that the overall SSIM similarity of the video segment to be detected is relatively low. If the overall structural similarity of the video segment to be detected is relatively low, it is determined that there is no door-blocking behavior, and enter Step 4; if the overall structural similarity of the video segment to be detected is relatively high, it is determined that there is door-blocking behavior, and output the video segment to be detected, and enter Step 4.

[0045] The SSIM similarity discrimination is adopted as an auxiliary discrimination method. When a large number of people continuously enter and exit the elevator, the difference between consecutive frame images is large, that is, the SSIM similarity is low; while when the door is blocked, the video is mostly in an approximately static state, and the difference between consecutive frame images is small, that is, the SSIM similarity is high. Therefore, using the SSIM similarity discrimination as an auxiliary discrimination method can better distinguish the door-blocking behavior from the behavior of a large number of people continuously entering and exiting the elevator.

[0046] Among them, further determine whether each video segment to be detected contains the door-blocking behavior, and output the video segment containing the door-blocking behavior, including the following steps: If the classification result output in is "may contain the door-blocking behavior", and it is determined that the overall structural similarity of the video segment to be detected is high, then it is determined that the video segment to be detected contains the door-blocking behavior, and the video segment to be detected is output. If the classification result output in is "may contain the door-blocking behavior", but it is determined that the overall structural similarity of the video segment to be detected is low, then it is determined that the video segment to be detected has no door-blocking behavior, and the video segment to be detected is not output. If the output classification result is "no door-blocking behavior", then it is determined that the video segment to be detected does not contain the door-blocking behavior, and the video segment to be detected is not output. This solution conducts multiple identifications and determinations on the door-blocking behavior in the elevator, can effectively identify and intercept the door-blocking segments, has a high recognition ability for the door-blocking behavior, reduces the misjudgment rate, and ensures the accuracy of the detection.

[0047] Step Four: Return to Step Two, and continue to detect whether the next video segment to be detected has the door-blocking behavior until all video segments to be detected are detected. The detection of the door-blocking behavior is completed without manual operation.

[0048] According to the output classification result and the calculated overall structural similarity of the video segment to be detected, the present invention further determines whether each video segment to be detected contains the door-blocking behavior, and outputs the video segment containing the door-blocking behavior.

[0049] Based on the above method for detecting the door-blocking behavior in the elevator, the present invention also proposes a computer-readable storage medium. Program codes are stored in the computer program, and the program codes are loaded and executed by a processor to implement the above method for detecting the door-blocking behavior in the elevator.

[0050] These program codes can also be loaded onto other programmable data processing devices, so that a series of operation steps are executed on other programmable devices to implement the above method for detecting the door-blocking behavior in the elevator.

[0051] The method proposed by the present invention realizes the accurate recognition of the door-blocking behavior in the elevator monitoring video, which has important significance for the safe operation, management and maintenance of the elevator.

[0052] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0053] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a means for implementing the functions specified in Figure 1 one or more of the flows Figure 1 or a plurality of flows and / or blocks

[0054] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction means that implements the functions specified in Figure 1 one or more of the flows Figure 1 or a plurality of flows and / or blocks

[0055] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one or more of the flows Figure 1 or a plurality of flows and / or blocks

[0056] Obviously, the above embodiments are merely examples given for clear illustration and are not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or variations can be made based on the above description. It is not necessary and impossible to exhaustively list all the implementation manners here. And the obvious changes or variations derived therefrom are still within the protection scope of the present invention.

Claims

1. A method for detecting the behavior of blocking the elevator door, characterized in that, It includes the following steps: S1. Equally divide the elevator monitoring video stream to be detected to obtain multiple video segments to be detected; S2. Obtain the pixel mosaic map of each video segment to be detected, and send it as input to a preset classification model. According to the classification result output by the classification model, if the output is likely to have a door-blocking behavior, go to step S3; if the output is no door-blocking behavior, go to step S4; S3. Calculate the structural similarity index between two adjacent frames in the video segment to be detected, and judge the overall structural similarity of the video segment to be detected according to the structural similarity index. If the overall similarity of the video segment to be detected is low, it is judged as no door-blocking behavior and go to step S4; if the overall similarity of the video segment to be detected is high, it is judged as having a door-blocking behavior and output the video segment to be detected; S4: Return to step S2 and continue to detect whether the next video segment to be detected has a door-blocking behavior until all video segments to be detected are detected; Among them, in S2, the method for obtaining the pixel mosaic map of each video segment to be detected specifically includes the following steps: Extract m frames of pictures from each video segment to be detected, and extract the nth row of pixels in the upper half of each frame of picture; set each nth row of pixels as a one-dimensional vector, and m one-dimensional vectors are extracted from each video segment to be detected; Arrange and splice the m one-dimensional vectors extracted from each video segment to be detected in the column direction in chronological order to form a pixel mosaic map with a width of m; The selection of the nth row of pixels follows the following steps: Start from the first row of pixels at the top of each frame of picture drawn, traverse each row of pixels from top to bottom, and calculate the gray-scale mean value of each row of pixels; Obtain the gray-scale mean value of the background pixel row of the picture, and compare the gray-scale mean value of each row of pixels with the gray-scale mean value of the background pixel row in turn; Set a threshold. When the first pixel row a whose difference from the gray-scale mean value of the background pixel row in the picture is greater than the threshold is searched, it is determined that the elevator door area starts from the a-th row of pixel positions, and the (a + 1)-th row of pixels is selected as the nth row of pixels.

2. The method for detecting the behavior of blocking the elevator door according to claim 1, characterized in that, In S1, "equally divide the elevator monitoring video stream to be detected" specifically includes: Obtain a sliding window with a length of s, and set the sliding step of the sliding window to q; Pass the elevator monitoring video stream to be detected through the sliding window, that is, intercept a video segment with a length of s from the elevator monitoring video stream to be detected every q seconds.

3. The method for detecting the behavior of blocking the elevator door according to claim 1, characterized in that, In S3, the calculation formula for the structural similarity index between two adjacent frames is: Among them, u X , u Y respectively represent the average gray values of two adjacent frames of images X and Y, σ X , σ Y respectively represent the standard deviations of the gray values of two adjacent frames of images X and Y, σ X 2 , σ Y 2 respectively represent the variances of the gray values of two adjacent frames of images X and Y, σ XY represents the covariance of the gray values of two adjacent frames of images X and Y, C1 and C2 are constants, and C1 = (K1 * L) 2 , C2 = (K2 * L) 2 , K1 = 0.01, K2 = 0.03, L = 255.

4. The method for detecting the behavior of blocking the elevator door according to claim 3, characterized in that, The method for judging the overall structural similarity of the video segment to be detected specifically includes: If the structural similarity index between two adjacent frames is detected to be greater than the threshold, start counting from these two adjacent frames. If the structural similarity indexes of L consecutive frames are greater than the threshold, it is determined that the overall similarity of the video segment to be detected is high; otherwise, it is determined that the overall similarity of the video segment to be detected is low.

5. The method for detecting the behavior of blocking the elevator door according to claim 1, characterized in that, The setting method of the classification model is: Collect and screen the monitoring videos of the elevator to obtain multiple video segments with door-blocking behaviors and without door-blocking behaviors; Successively extract frames from all video segments, extract a row of pixels at the same position in each frame, and splice the multiple extracted pixel rows in sequence to obtain a pixel splicing map for each video segment; Divide the multiple pixel splicing maps into two categories: with door blocking behavior and without door blocking behavior, extract the image features in the two categories of pixel splicing maps, and train the image features to establish a classification model.

6. The method for detecting the behavior of blocking the elevator door according to claim 5, characterized in that, Send the two categories of pixel splicing maps to the MobileNetV3 convolutional neural network for image feature extraction.

7. The method for detecting the behavior of blocking the elevator door according to claim 6, characterized in that, The establishment of the classification model also includes: Divide the two categories of pixel splicing maps with door blocking behavior and without door blocking behavior into a training set and a validation set according to a preset ratio; Input the training set into the MobileNetV3 convolutional neural network, use the MobileNetV3 convolutional neural network to extract the image features of the training set, and train the image features to establish a classification model; Use the validation set to verify the accuracy of the classification model.

8. A computer-readable storage medium, on which a computer program is stored, characterized in that, The computer program stores program code, which is loaded and executed by a processor to implement the elevator door blocking behavior detection method according to any one of claims 1 to 7.

Images