Elevator monitoring video still image recognition system and method based on SSIM algorithm

Abstract

The present invention discloses a still picture recognition system and method for elevator monitoring videos based on the SSIM algorithm, comprising: a video acquisition module for collecting monitoring video data in an elevator and transmitting the monitoring video data to a preprocessing module; a preprocessing module for preprocessing the monitoring video data to obtain target video data; a video processing module for processing the target video data using the SSIM algorithm to obtain the inter-frame similarity of the target video data; a still picture recognition module for setting a corresponding threshold, comparing the threshold with the inter-frame similarity, and determining whether a still picture is obtained; and a recording and output module for recording still picture information corresponding to the still picture and outputting the still picture information to a display device. The present invention can simultaneously save dynamic and static pictures in elevator monitoring videos, ensure the integrity of the monitoring pictures, and improve the accuracy of video monitoring. Unlike existing AI algorithms, the SSIM algorithm used in the present invention has low computational complexity and low cost.

Description

Technical Field

[0001] The present invention relates to the technical field of digital image processing and computer vision, and in particular to a system and method for recognizing still images of elevator monitoring videos based on an SSIM algorithm. Background Art

[0002] Elevators, as an indispensable means of transportation in modern buildings, have always garnered significant attention for their safety. Elevator surveillance video utilizes video acquisition technology, installing high-definition cameras inside elevator cars and in key operating areas. By recording and transmitting real-time video signals during elevator operation, comprehensive monitoring and recording of elevator operation status is achieved. Elevator surveillance video utilizes high-definition cameras to provide high-definition monitoring of the elevator interior. This data is transmitted to a monitoring center via network communication technology, enabling remote, real-time monitoring of the elevator. Algorithms also analyze the video stream in real time, identifying specific scenarios and behaviors, such as entrapment and abnormal behavior, and providing appropriate feedback. To ensure the long-term preservation of important footage, non-essential footage can be intelligently filtered, saving storage space. Therefore, elevator surveillance video is a crucial component of elevator safety and management, providing a strong guarantee for its safe operation. The SSIM algorithm, an image quality assessment method that mimics the characteristics of the human visual system, comprehensively considers similarities in brightness, contrast, and structure to more accurately reflect the similarity between video images. Consequently, the SSIM algorithm can be applied to still image recognition in elevator surveillance video.

[0003] Traditional elevator monitoring systems rely on manual monitoring, but this approach often suffers from low efficiency, fatigue, and a high false alarm rate. Furthermore, to conserve storage space and bandwidth, existing elevator monitoring systems typically only save dynamic images, failing to save static images. This prevents the integrity of the monitoring images and the completeness of the surveillance video, resulting in poor video monitoring accuracy. While existing AI algorithms can also save static images of surveillance videos and detect the presence of people or objects in the elevator in real time, this approach is computationally intensive and costly.

[0004] Therefore, how to solve the problems that traditional elevator monitoring systems cannot save static images, cannot ensure the integrity of monitoring images and monitoring videos, have poor accuracy of video monitoring, and the existing algorithms have large computational complexity and high cost, has become a problem that needs to be solved urgently. Summary of the Invention

[0005] In order to solve the above problems, the present invention proposes a still picture recognition system and method for elevator monitoring video based on the SSIM algorithm, which collects monitoring video data in the elevator and transmits the monitoring video data to a preprocessing module; preprocesses the monitoring video data to obtain preprocessed monitoring video data, i.e., target video data; processes the target video data through the SSIM algorithm to obtain the inter-frame similarity of the target video data; sets a corresponding threshold, compares the threshold with the inter-frame similarity, and determines to obtain a still picture; records the still picture information corresponding to the still picture, and outputs the still picture information to a display device, so that not only the dynamic pictures in the elevator monitoring video can be saved, but also the static pictures can be saved, thereby ensuring the integrity of the elevator monitoring picture and the integrity of the monitoring video, and improving the accuracy of elevator video monitoring. At the same time, unlike the existing AI algorithm, the SSIM algorithm used in the present invention has a small amount of calculation and low cost, and is more suitable for actual elevator monitoring video systems.

[0006] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0007] The present invention provides an elevator monitoring video still image recognition system based on the SSIM algorithm, comprising:

[0008] A video acquisition module is used to collect monitoring video data in the elevator and transmit the monitoring video data to the preprocessing module;

[0009] A preprocessing module is used to preprocess the monitoring video data to obtain preprocessed monitoring video data, i.e., target video data;

[0010] A video processing module is used to process the target video data using an SSIM algorithm to obtain inter-frame similarity of the target video data;

[0011] A still picture recognition module is used to set a corresponding threshold, compare the threshold with the inter-frame similarity, and determine whether a still picture is obtained;

[0012] The recording and outputting module is used to record the still picture information corresponding to the still picture and output the still picture information to a display device.

[0013] A further improvement of the present invention is that the video acquisition module includes:

[0014] An acquisition unit, configured to collect monitoring video data inside the elevator in real time and acquire the monitoring video data;

[0015] A transmission unit is used to transmit the monitoring video data to the preprocessing module.

[0016] A further improvement of the present invention is that the preprocessing module is used to preprocess the monitoring video data using the following video preprocessing algorithm to obtain preprocessed monitoring video data, i.e., target video data:

[0017] Convert the target video data from RGB channels to HSV channels to obtain H channel data, S channel data, and V channel data respectively;

[0018] The V channel data is enhanced twice, including:

[0019] Perform a first enhancement process on the V channel data to obtain first enhanced channel data. The expression of the first enhanced channel data is:

[0020]

[0021] Where V1(x,y) represents the pixel value of the first enhanced channel data at the pixel point (x,y); v(x,y) represents the pixel value of the V channel data at the pixel point (x,y); max represents the maximum value;

[0022] Perform a second enhancement process on the first enhanced channel data to obtain second enhanced channel data. The expression of the second enhanced channel data is:

[0023] V2(x,y)=(0.013Y+0.015)[V1(x,y)] 3 +(1.093Y+0.836)[V1(x,y)] 2 -(3.079Y

[0024] -1.734)V1(x,y)+(2.978Y-2.606)

[0025] Where V2(x,y) represents the pixel value of the second enhanced channel data at the pixel point (x,y); V1(x,y) represents the pixel value of the first enhanced channel data at the pixel point (x,y); Y represents the enhanced distinction threshold;

[0026] The second enhanced channel data is the V channel data after two enhancement processes. The V channel data after two enhancement processes are merged with the H channel data and the S channel data to obtain pre-processed monitoring video data, namely the target video data.

[0027] A further improvement of the present invention is that the video processing module is used to process the target video data using the following SSIM similarity algorithm to obtain the inter-frame similarity of the target video data:

[0028] Selecting a local window and sliding the local window on the video frame of the target video data;

[0029] Calculate the brightness mean and contrast variance of the pixels in each local window, as well as the structural covariance with the pixels in other local windows, to obtain the statistical characteristics of the local window;

[0030] According to the statistical features, the SSIM values ​​of adjacent video frames of the target video data are calculated, and the expression of the SSIM value is:

[0031]

[0032] Where m and n represent adjacent video frames in the target video data; μ m 、μ n Respectively represent the brightness mean of pixels in adjacent video frames m and n in the local window; σ m 2 , σ n 2 Respectively represent the contrast variance of pixels in adjacent video frames m and n in the local window; σ mn Represents the structural covariance of pixels of adjacent video frames m and n in the local window; a1 and a2 are constants, and a1≠0 and a2≠0 to avoid the denominator being zero;

[0033] After obtaining the SSIM values ​​of the adjacent video frames of the target video data, performing a weighted average of the SSIM values ​​of all windows in the video frame to obtain a global SSIM value, i.e., the inter-frame similarity of the target video data, wherein the value of the inter-frame similarity ranges from 0 to 1, and a larger value of the inter-frame similarity indicates a higher similarity between the adjacent video frames of the target video data;

[0034] The brightness, contrast and structure similarities of adjacent video frames of the target video data are compared by using the SSIM algorithm to obtain the inter-frame similarity of the target video data.

[0035] A further improvement of the present invention is that the still picture recognition module includes:

[0036] A setting unit, configured to set a corresponding threshold value according to the monitoring requirements of the elevator;

[0037] A receiving unit, configured to receive the inter-frame similarity of the target video data;

[0038] A comparison and identification unit is used to compare the threshold with the inter-frame similarity, count the number of consecutive video frames whose inter-frame similarity exceeds the threshold, and if the number of consecutive video frames meets a preset number condition, determine that the consecutive video frames correspond to a still picture.

[0039] A further improvement of the present invention is that: the preset number condition is that the number of the continuous video frames is greater than or equal to a preset still number.

[0040] A further improvement of the present invention is that the record output module includes:

[0041] a marking unit, configured to mark the continuous video frames to obtain marked video frames;

[0042] a recording unit, configured to record still image information corresponding to the marked video frame;

[0043] An output unit is used to output the still picture information to the display device.

[0044] The present invention provides a method for identifying still images of elevator monitoring videos based on the SSIM algorithm, comprising:

[0045] Collecting surveillance video data in the elevator and transmitting the surveillance video data to the pre-processing module;

[0046] Preprocessing the surveillance video data to obtain preprocessed surveillance video data, i.e., target video data;

[0047] The target video data is processed by the SSIM algorithm to obtain the inter-frame similarity of the target video data;

[0048] Setting a corresponding threshold, comparing the threshold with the inter-frame similarity, and determining whether a still picture is obtained;

[0049] Still picture information corresponding to the still picture is recorded, and the still picture information is output to a display device.

[0050] The beneficial effects of the present invention are: the present invention proposes a still picture recognition system and method for elevator monitoring video based on the SSIM algorithm, which collects monitoring video data in the elevator and transmits the monitoring video data to a preprocessing module; preprocesses the monitoring video data to obtain preprocessed monitoring video data, that is, target video data; processes the target video data through the SSIM algorithm to obtain the inter-frame similarity of the target video data; sets a corresponding threshold, compares the threshold with the inter-frame similarity, and determines to obtain a still picture; records the still picture information corresponding to the still picture, and outputs the still picture information to a display device, so that not only the dynamic picture in the elevator monitoring video can be saved, but also the static picture can be saved, thereby ensuring the integrity of the elevator monitoring picture and the integrity of the monitoring video, and improving the accuracy of elevator video monitoring. At the same time, unlike the existing AI algorithm, the SSIM algorithm used in the present invention has a small amount of calculation and low cost, and is more suitable for actual elevator monitoring video systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A schematic diagram of the structure of a still image recognition system for elevator monitoring video based on the SSIM algorithm provided by an embodiment of the present invention;

[0052] Figure 2 A schematic diagram of the structure of a video acquisition module of an elevator monitoring video still image recognition system based on the SSIM algorithm provided in an embodiment of the present invention;

[0053] Figure 3 A schematic diagram of the structure of a still picture recognition module of an elevator monitoring video still picture recognition system based on the SSIM algorithm provided in an embodiment of the present invention;

[0054] Figure 4 A schematic diagram of the structure of a recording output module of an elevator monitoring video still image recognition system based on the SSIM algorithm provided in an embodiment of the present invention;

[0055] Figure 5 The present invention provides a flow chart of a method for still image recognition in elevator surveillance video based on the SSIM algorithm. DETAILED DESCRIPTION

[0056] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only intended to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0057] See also Figure 1 , a still image recognition system for elevator surveillance video based on the SSIM algorithm of this embodiment includes:

[0058] The video acquisition module is used to collect the monitoring video data in the elevator and transmit the monitoring video data to the preprocessing module;

[0059] The preprocessing module is used to preprocess the monitoring video data to obtain the preprocessed monitoring video data, i.e., the target video data;

[0060] The video processing module is used to process the target video data using the SSIM algorithm to obtain the inter-frame similarity of the target video data;

[0061] A still picture recognition module is used to set a corresponding threshold, compare the threshold with the inter-frame similarity, and determine whether a still picture is obtained;

[0062] The recording and outputting module is used to record the still picture information corresponding to the still picture and output the still picture information to the display device.

[0063] In practical applications, SSIM (Structural Similarity Index Measure) is a metric used to measure the similarity between two images. This algorithm evaluates image similarity by comparing three key features: brightness, contrast, and structure. The SSIM algorithm is based on three key image characteristics: brightness (measured as the average grayscale value, obtained by calculating the average value of all pixel values ​​in the image); contrast (measured as the standard deviation of grayscale values, reflecting the degree of brightness changes in the image); and structure (compared to the normalized correlation of pixel values ​​in the image, i.e., the image's structural information). The SSIM algorithm calculates the similarity between the two images in terms of brightness, contrast, and structure, and then weights and multiplies these three similarity values ​​to obtain the final SSIM value. SSIM values ​​range from -1 to 1, with larger values ​​indicating greater similarity between the two images.

[0064] Among them, the SSIM algorithm has a wide range of applications in the field of image processing, including: image quality assessment: used to evaluate the quality loss after image compression, transmission or restoration; image denoising: as a performance evaluation indicator of the denoising algorithm; image enhancement: in the image enhancement process, the enhancement effect is evaluated by comparing the images before and after enhancement; image reconstruction: in the image reconstruction task, it is used as a loss function to guide the reconstruction process. The SSIM algorithm is an effective image similarity measurement method, which evaluates the similarity between images by comparing the three key features of image brightness, contrast and structure. The algorithm has broad application prospects and important research value in the field of image processing. Therefore, the present invention applies the SSIM algorithm to the elevator monitoring video system, which can not only save the dynamic images in the elevator monitoring video, but also save the static images, thereby ensuring the integrity of the elevator monitoring images and the integrity of the monitoring video, and improving the accuracy of elevator video monitoring. At the same time, unlike the existing AI algorithm, the SSIM algorithm used in the present invention has a small amount of computation and low cost, and is more suitable for actual elevator monitoring video systems.

[0065] Among them, see Figure 2 , video acquisition module, including:

[0066] An acquisition unit, used to collect monitoring video data inside the elevator in real time and obtain the monitoring video data;

[0067] The transmission unit is used to transmit the monitoring video data to the pre-processing module.

[0068] It is understandable that the video acquisition module can input the collected surveillance video data inside the elevator into the subsequent processing module in frames, so as to facilitate the subsequent corresponding processing of the video frames in the surveillance video data through the SSIM algorithm to obtain the inter-frame similarity of the surveillance video data.

[0069] Through the above steps, the processing efficiency and accuracy of elevator monitoring can be improved, the situation inside the elevator can be monitored in real time, and the automation level of elevator monitoring video can be improved.

[0070] The preprocessing module is used to preprocess the surveillance video data using the following video preprocessing algorithm to obtain preprocessed surveillance video data, i.e., target video data:

[0071] Convert the target video data from RGB channels to HSV channels to obtain H channel data, S channel data, and V channel data respectively;

[0072] The V channel data is enhanced twice, including:

[0073] Perform the first enhancement processing on the V channel data to obtain the first enhanced channel data. The expression of the first enhanced channel data is:

[0074]

[0075] Where V1(x,y) represents the pixel value of the first enhanced channel data at the pixel point (x,y); v(x,y) represents the pixel value of the V channel data at the pixel point (x,y); max represents the maximum value;

[0076] The first enhanced channel data is enhanced for the second time to obtain the second enhanced channel data. The expression of the second enhanced channel data is:

[0077] V2(x,y)=(0.013Y+0.015)[V1(x,y)] 3 +(1.093Y+0.836)[V1(x,y)] 2 -(3.079Y

[0078] -1.734)V1(x,y)+(2.978Y-2.606)

[0079] Where V2(x,y) represents the pixel value of the second enhanced channel data at the pixel point (x,y); V1(x,y) represents the pixel value of the first enhanced channel data at the pixel point (x,y); Y represents the enhanced distinction threshold;

[0080] The second enhanced channel data is the V channel data after the two enhancement processes. The V channel data after the two enhancement processes are merged with the H channel data and the S channel data to obtain the preprocessed monitoring video data, that is, the target video data.

[0081] In practical applications, preprocessing the acquired surveillance video data can significantly improve its quality, thereby improving the accuracy and efficiency of subsequent processing. At the same time, it can preserve or restore detailed information in the surveillance video, making it clearer and improving its visual quality. It can also adjust parameters such as color saturation and brightness to make the colors of the surveillance video more vivid and lively.

[0082] The video processing module is used to process the target video data using the following SSIM similarity algorithm to obtain the inter-frame similarity of the target video data:

[0083] Select a local window and slide the local window on the video frame of the target video data;

[0084] By calculating the brightness mean and contrast variance of the pixels in each local window, as well as the structural covariance with the pixels in other local windows, the statistical characteristics of the local window can be obtained;

[0085] According to the statistical characteristics, the SSIM values ​​of adjacent video frames of the target video data are calculated. The expression of the SSIM value is:

[0086]

[0087] Where m and n represent adjacent video frames in the target video data; μ m 、μ n Respectively represent the brightness mean of pixels in adjacent video frames m and n in the local window; σ m 2 , σ n 2 Respectively represent the contrast variance of pixels in adjacent video frames m and n in the local window; σ mn Represents the structural covariance of pixels of adjacent video frames m and n in the local window; a1 and a2 are constants, and a1≠0 and a2≠0 to avoid the denominator being zero;

[0088] After obtaining the SSIM values ​​of adjacent video frames of the target video data, the SSIM values ​​of all windows in the video frame are weighted averaged to obtain the global SSIM value, that is, the inter-frame similarity of the target video data. The value range of the inter-frame similarity is between 0 and 1. The larger the inter-frame similarity value, the higher the similarity of adjacent video frames of the target video data.

[0089] The SSIM algorithm is used to compare the brightness, contrast and structure similarity of adjacent video frames of the target video data to obtain the inter-frame similarity of the target video data.

[0090] Through the above steps, the accuracy of elevator video monitoring can be improved. Different from the existing AI algorithm, the SSIM algorithm used in the present invention has a smaller computational complexity and lower cost, and is more suitable for actual elevator monitoring video systems.

[0091] Among them, see Figure 3 , the still picture recognition module includes:

[0092] A setting unit, used to set corresponding thresholds according to the monitoring requirements of the elevator;

[0093] A receiving unit, configured to receive inter-frame similarity of target video data;

[0094] The comparison and identification unit is used to compare the threshold with the inter-frame similarity, count the number of consecutive video frames whose inter-frame similarity exceeds the threshold, and determine that the consecutive video frames correspond to still images if the number of consecutive video frames meets a preset number condition.

[0095] It should be noted that the preset number condition is that the number of consecutive video frames is greater than or equal to the preset still number.

[0096] It is understandable that when setting the corresponding threshold, factors such as the clarity, frame rate and definition of still images of the actual elevator monitoring video need to be considered to meet the relevant monitoring needs of the actual elevator. After determining the required threshold, the inter-frame similarity obtained by the video processing module can be compared with the threshold. Among them, the preset still number refers to the number of continuous video frames for judging still images preset in advance by the staff. When the inter-frame similarity of multiple consecutive frames is greater than the threshold, and the number of continuous video frames is greater than the preset still number, it can be judged that these continuous video frames correspond to still images.

[0097] Through the above steps, not only the dynamic images in the elevator monitoring video can be saved, but also the static images can be saved, which ensures the integrity of the elevator monitoring images and the integrity of the monitoring video, and improves the accuracy of elevator video monitoring. At the same time, unlike the existing AI algorithm, the SSIM algorithm used in the present invention has a small amount of calculation and low cost, and is more suitable for actual elevator monitoring video systems.

[0098] Among them, see Figure 4 , record output module, including:

[0099] a marking unit, configured to mark continuous video frames to obtain marked video frames;

[0100] a recording unit, configured to record still image information corresponding to the marked video frame;

[0101] The output unit is used to output the still picture information to a display device.

[0102] In practical applications, continuous video frames determined in the monitoring video can be automatically marked to obtain corresponding marked video frames. These marked video frames correspond to still images, and the corresponding still image information can be recorded, such as still time, position, etc. After obtaining the still image information, the still image information can be output to the display device corresponding to the elevator monitoring, so that the management personnel can view and process it in time and take corresponding measures, thereby improving the real-time performance of elevator monitoring and facilitating the staff to provide feedback and handle the corresponding situations in time.

[0103] See also Figure 5 The present invention provides a method for recognizing still images of elevator monitoring videos based on the SSIM algorithm, comprising:

[0104] Collect surveillance video data in the elevator and transmit the surveillance video data to the pre-processing module;

[0105] Preprocessing the surveillance video data to obtain preprocessed surveillance video data, i.e., target video data;

[0106] The target video data is processed by the SSIM algorithm to obtain the inter-frame similarity of the target video data;

[0107] Set a corresponding threshold, compare the threshold with the inter-frame similarity, and determine whether a still image is obtained;

[0108] Still picture information corresponding to the still picture is recorded, and the still picture information is output to a display device.

[0109] Through the introduction of the above embodiments, the present invention can collect monitoring video data in the elevator through a still picture recognition system and method of elevator monitoring video based on the SSIM algorithm, and transmit the monitoring video data to a preprocessing module; preprocess the monitoring video data to obtain preprocessed monitoring video data, that is, target video data; process the target video data through the SSIM algorithm to obtain the inter-frame similarity of the target video data; set a corresponding threshold, compare the threshold with the inter-frame similarity, and determine to obtain a still picture; record the still picture information corresponding to the still picture, and output the still picture information to a display device, so that not only the dynamic picture in the elevator monitoring video can be saved, but also the static picture can be saved, thereby ensuring the integrity of the elevator monitoring picture and the integrity of the monitoring video, and improving the accuracy of elevator video monitoring. At the same time, unlike the existing AI algorithm, the SSIM algorithm used in the present invention has a small amount of calculation and low cost, and is more suitable for actual elevator monitoring video systems.

[0110] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art and, unless defined similarly as herein, will not be interpreted in an idealized or overly formal sense.

[0111] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is 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 in the scope of protection of the present invention.

Claims

1. An elevator surveillance video still image recognition system based on the SSIM algorithm, characterized by: include: A video acquisition module is used to collect monitoring video data in the elevator and transmit the monitoring video data to the preprocessing module; A preprocessing module is used to preprocess the monitoring video data to obtain preprocessed monitoring video data, i.e., target video data; A video processing module is used to process the target video data using an SSIM algorithm to obtain inter-frame similarity of the target video data; A still picture recognition module, configured to set a corresponding threshold, compare the threshold with the inter-frame similarity, and determine whether a still picture is obtained; a recording and outputting module, configured to record still picture information corresponding to the still picture, and output the still picture information to a display device; The video acquisition module includes: An acquisition unit, configured to collect monitoring video data inside the elevator in real time and acquire the monitoring video data; A transmission unit, configured to transmit the monitoring video data to the pre-processing module; The preprocessing module is used to preprocess the monitoring video data using the following video preprocessing algorithm to obtain preprocessed monitoring video data, i.e., target video data: Convert the target video data from RGB channels to HSV channels to obtain H channel data, S channel data, and V channel data respectively; The V channel data is enhanced twice, including: Perform a first enhancement process on the V channel data to obtain first enhanced channel data. The expression of the first enhanced channel data is: Where V1(x,y) represents the pixel value of the first enhanced channel data at the pixel point (x,y); v(x,y) represents the pixel value of the V channel data at the pixel point (x,y); max represents the maximum value; Perform a second enhancement process on the first enhanced channel data to obtain second enhanced channel data. The expression of the second enhanced channel data is: V2(x,y)=(0.013Y+0.015)[V1(x,y)] 3 +(1.093Y+0.836)[V1(x,y)] 2 +(3.079Y -1.734)V1(x,y)+(2.978Y-2.606) Where v2(x,y) represents the pixel value of the second enhanced channel data at the pixel point (x,y); V1(x,y) represents the pixel value of the first enhanced channel data at the pixel point (x,y); Y represents the enhanced distinction threshold; The second enhanced channel data is the V channel data after the two enhancement processes, and the V channel data after the two enhancement processes are merged with the H channel data and the S channel data to obtain the pre-processed monitoring video data, i.e., the target video data; The video processing module is configured to process the target video data using the following SSIM similarity algorithm to obtain inter-frame similarity of the target video data: Selecting a local window and sliding the local window on the video frame of the target video data; Calculate the brightness mean and contrast variance of the pixels in each local window, as well as the structural covariance with the pixels in other local windows, to obtain the statistical characteristics of the local window; According to the statistical features, the SSIM values ​​of adjacent video frames of the target video data are calculated, and the expression of the SSIM value is: Where m and n represent adjacent video frames in the target video data; μ m 、μ n Respectively represent the brightness mean of pixels in adjacent video frames m and n in the local window; σ m 2 , σ n 2 Respectively represent the contrast variance of pixels in adjacent video frames m and n in the local window; σ mn Represents the structural covariance of pixels of adjacent video frames m and n in the local window; a1 and a2 are constants, and a1≠0 and a2≠0 to avoid the denominator being zero; After obtaining the SSIM values ​​of the adjacent video frames of the target video data, performing a weighted average of the SSIM values ​​of all windows in the video frame to obtain a global SSIM value, i.e., the inter-frame similarity of the target video data, wherein the value of the inter-frame similarity ranges from 0 to 1, and a larger value of the inter-frame similarity indicates a higher similarity between the adjacent video frames of the target video data; Comparing the brightness, contrast, and structure of adjacent video frames of the target video data using the SSIM algorithm to obtain inter-frame similarity of the target video data; The still picture recognition module includes: A setting unit, configured to set a corresponding threshold value according to the monitoring requirements of the elevator; A receiving unit, configured to receive the inter-frame similarity of the target video data; a comparison and identification unit, configured to compare the threshold with the inter-frame similarity, count the number of consecutive video frames whose inter-frame similarity exceeds the threshold, and determine that the consecutive video frames correspond to still images if the number of consecutive video frames meets a preset number condition; The preset number condition is that the number of the continuous video frames is greater than or equal to a preset still number; the recording output module includes: a marking unit, configured to mark the continuous video frames to obtain marked video frames; a recording unit, configured to record still image information corresponding to the marked video frame; An output unit is used to output the still picture information to the display device.

2. The recognition method of the elevator monitoring video still image recognition system based on the SSIM algorithm according to claim 1 is characterized in that: include: Collecting surveillance video data in the elevator and transmitting the surveillance video data to the pre-processing module; Preprocessing the surveillance video data to obtain preprocessed surveillance video data, i.e., target video data; The target video data is processed by the SSIM algorithm to obtain the inter-frame similarity of the target video data; a corresponding threshold is set, the threshold is compared with the inter-frame similarity, and a still picture is determined; still picture information corresponding to the still picture is recorded, and the still picture information is output to a display device.

Images