A method for monitoring the physical security environment of an office area

Abstract

The present invention discloses a method for monitoring the physical security environment of an office area. In order to solve the problem that the physical security environment classification of an office area is not objective and accurate enough, the present invention discloses the use of deep learning to build an office area image processing model and train and learn the office area image processing model. Through the training of the office area image processing model, different physical security environment levels are classified according to the office environment image of the area to be detected, thereby improving the processing effect of the office area image. The physical security environment alarm module realizes the objective and accurate classification alarm triggering of the physical security environment based on the office area image processing model. The present invention reduces the cost and complexity of manpower monitoring of privacy and confidential office areas through the built deep learning model, realizes accurate and efficient monitoring levels, realizes mechanized and accurate security level monitoring for different special confidential office environments, improves the security monitoring level of confidential areas, and thus protects the business secrets of companies.

Description

Technical Field

[0001] The present invention belongs to the technical field of work environment safety monitoring, and in particular relates to a method for monitoring the physical safety environment of an office area. Background Art

[0002] With the development of smart cities and the Internet of Things, there are more and more office buildings and office areas for centralized office work. These places are densely populated with people and have very complex composition. How to ensure the safety of these areas is extremely important. With the development of automation and digitalization, many office buildings and office areas have installed monitoring systems for fire protection or security purposes. However, most of these systems only use video images for monitoring, and the operating efficiency is not high, making it difficult to provide all-round and high-standard protection for the above-mentioned areas.

[0003] Furthermore, the use of security systems is becoming more and more widespread, especially for many office areas where there will be no blind spot security monitoring. Although this has greatly increased the security effect, it cannot protect the privacy of relevant staff well. Nowadays, many office areas are equipped with surveillance cameras, which have very good protection and traceability for the loss of employee workstation items or related data leakage, but it is likely that the privacy of employees will not be protected accordingly, which is not conducive to protecting the self-esteem of employees and will also increase the work pressure of employees. However, if more stringent security monitoring is not adopted, it will bring greater security risks to the company. The existing ones are generally equipped with a camera control and shutdown module. After the information authentication, the employee can independently control the opening or closing of the camera module, but this design is easy to make the camera closed area uncontrollable, and there are control contradictions that employees can turn off the camera or repeatedly turn it on or off in the same area.

[0004] Through the above analysis, the existing technology has the following problems to be solved:

[0005] How to combine office area monitoring images and sensor device parameters: It is necessary to develop new physical security environment monitoring methods for office areas that can make full use of multimodal parameters of the office environment to adapt to complex physically separated office areas, thereby improving the accuracy and security of automated office area monitoring.

[0006] The processing of images of physically separated areas is to achieve the privacy protection of relevant staff members, and the image data is optimized to a certain extent. It is necessary to design a specific office environment area processing algorithm based on machine learning algorithm to improve the confidentiality of the image.

[0007] Therefore, it is necessary to redesign a method for monitoring the physical security environment of office areas in response to existing problems. Summary of the invention

[0008] In order to solve the above technical problems, the present invention proposes a method for monitoring the physical security environment of an office area.

[0009] In a first aspect of the present invention, a method for monitoring the physical security environment of an office area is provided, the method comprising:

[0010] Obtain the physical separation area image and corresponding parameters, and obtain the current office physical security environment level;

[0011] Constructing an office area image processing model according to the physical separation area image and corresponding parameters and the current office physical security environment level;

[0012] Acquire a current physical separation area image and corresponding parameters, process the current physical separation area image and corresponding parameters according to an office area image processing model, and generate a current office physical security environment level;

[0013] The current office physical security environment level is obtained, and alarm processing is performed according to the current office physical security environment level.

[0014] Furthermore, the office area image processing model is constructed using machine learning or deep learning methods.

[0015] Furthermore, the physical separation area image and corresponding parameters include:

[0016] The images of the physical separation area captured by the video acquisition device, the average of the parameters collected by the sensors of the physical separation area and the real-time parameters of the area given by the corresponding electronic equipment of the separation area, and the number of times the security alarm of the electronic equipment is triggered.

[0017] A physical security environment monitoring system for an office area is also provided, wherein the system comprises:

[0018] An office environment image acquisition module, which acquires a physical separation area image and corresponding parameters, and acquires the current office physical security environment level according to the physical separation area image and corresponding parameters;

[0019] An office area image processing module, which constructs an office area image processing model according to the physical separation area image and corresponding parameters and the current office physical security environment level;

[0020] The office environment image acquisition module is also used to receive the current physical separation area image and corresponding parameters, and is connected to the office area image processing module, and processes the current physical separation area image and corresponding parameters according to the office area image processing model in the office area image processing module to generate the current office physical security environment level;

[0021] The physical security environment alarm module receives the current office physical security environment level and issues an alarm level prompt according to the current office physical security environment level.

[0022] Furthermore, the office area image processing model is constructed using machine learning or deep learning methods.

[0023] Furthermore, the physical separation area image and corresponding parameters include:

[0024] The images of the physical separation area captured by the video acquisition device, the average of the parameters collected by the sensors of the physical separation area and the real-time parameters of the area given by the corresponding electronic equipment of the separation area, and the number of times the security alarm of the electronic equipment is triggered.

[0025] The physical security environment monitoring of the office area carried out by this method is safer and more accurate. On the one hand, it can improve the accuracy of the security level alarm of the office area monitoring. On the other hand, it integrates the image data of the office area and the parameters obtained by sensors and electronic devices online. It can not only obtain the specific disaster alarm level, but also monitor whether employees have issued an alarm for the leakage of commercial secrets and provide an alarm level. Finally, the acquired image data is also processed by the improved ResNet50 model, and the continuously changing image data is more suitable for processing in the specific application scenario of the office environment.

[0026] More embodiments and improved effects of the present invention will be further introduced in conjunction with the drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a flow chart of a method for monitoring the physical security environment of an office area of ​​the present invention;

[0028] Figure 2 It is a schematic diagram of a physical security environment monitoring system for an office area of ​​the present invention;

[0029] Figure 3 is a schematic diagram specifically showing the number of physically separated regions in an embodiment of the present invention;

[0030] Figure 4 It is a schematic diagram of the structure of an electronic device for implementing the method of the present invention according to an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The invention is further described below in conjunction with the accompanying drawings and specific implementation methods.

[0032] In order to solve the above technical problems, the present invention proposes a method for monitoring the physical security environment of an office area.

[0033] In a first aspect of the present invention, a method for monitoring the physical security environment of an office area is provided, the method comprising:

[0034] Obtain the physical separation area image and corresponding parameters, and obtain the current office physical security environment level;

[0035] Constructing an office area image processing model according to the physical separation area image and corresponding parameters and the current office physical security environment level;

[0036] Acquire a current physical separation area image and corresponding parameters, process the current physical separation area image and corresponding parameters according to an office area image processing model, and generate a current office physical security environment level;

[0037] The current office physical security environment level is obtained, and alarm processing is performed according to the current office physical security environment level.

[0038] Furthermore, the office area image processing model is constructed using machine learning or deep learning methods.

[0039] Furthermore, the physical separation area image and corresponding parameters include:

[0040] The images of the physical separation area captured by the video acquisition device, the average of the parameters collected by the sensors of the physical separation area and the real-time parameters of the area given by the corresponding electronic equipment of the separation area, and the number of times the security alarm of the electronic equipment is triggered.

[0041] In this embodiment, the average value between the parameters collected by the sensor of the physical separation area and the real-time parameters of the area given by the corresponding electronic device of the separation area are specifically weather and climate parameters, such as temperature and air humidity. If the temperature collected by the sensor is 25°C and the air humidity collected by the sensor is 30%, the temperature and air humidity automatically collected by the corresponding electronic device of the separation area can be 27°C and 32% respectively after the mobile phone of the staff is connected to the office network or the computer is connected to the office network. The average value is 26°C and 31%. The threshold value of the number of times the electronic device security alarm is triggered is set to 2 times. The set number of times is determined by the number of alarms after the office area personnel use the prohibited software for more than 20 seconds. At this time, the corresponding corresponding parameter characteristic value is (26,31,2).

[0042] Generally, the update rate attenuation methods commonly used in training conventional ResNet models include adjusting the update rate at equal intervals and adjusting the update rate at exponential decay.

[0043] The method of adjusting the update rate at equal intervals. After the model is trained for a certain number of iterations, the update rate decreases by a certain percentage. This attenuation strategy is prone to cause oscillation when changing the update rate due to the large change in the update rate attenuation, and it cannot converge to the optimal value.

[0044] Exponential decay method to adjust the update rate. The model dynamically changes the update rate, and the update frequency of the update rate is determined by the number of times. When the decay rate is small, as the number of times gradually increases, the change in the update rate becomes smaller, which can easily lead to the model converging too quickly.

[0045] For images of specific office environment areas, the applicant discovered through multiple optimization and updates of the image feature vector that changing the number of physically separated areas would have a significant impact on the model.

[0046] The physical separation area images obtained in this embodiment are processed using the SM-ResNet50 model. First, the SM-ResNet50 model is selected, the fully connected layer is removed, and features are extracted for the 1200 physical separation area images in the physical separation area image data set. After feature extraction, each physical separation area image is converted into a 1024-dimensional feature vector. The SM-ResNet50 model is generated by training the data set.

[0047] Furthermore, the improved regional image processing SM-ResNet50 model is used to process the physically separated regional images. The specific update rate σ t The formula is:

[0048]

[0049] Where: i is the change value; is the maximum value of the update rate, is the minimum value of the update rate, D cur is the current number of updates, D i is the total number of changes, The number of physical areas that the office area is physically separated into according to the number of times it has been updated.

[0050] In the office area on the first floor, the physical separation area is generally selected to be 8-16 rooms, such as Figure 4 The middle office area is physically divided into 10 areas. In this embodiment, the total number of updates is selected as 2500 times, the update rate is selected as between 0.05-0.1, and the change value is selected as between 0.2-0.5.

[0051] Furthermore, the deep learning method uses an improved GAE function to train the model:

[0052]

[0053] In the formula, G(X) is the activation function value, is the number of physical areas divided by the physical separation area of ​​the office area according to the number of times it is updated, and x is the input feature value, which is obtained from the physical separation area image and corresponding parameters.

[0054] A physical security environment monitoring system for an office area is also provided, wherein the system comprises:

[0055] An office environment image acquisition module, which acquires a physical separation area image and corresponding parameters, and acquires the current office physical security environment level according to the physical separation area image and corresponding parameters;

[0056] An office area image processing module, which constructs an office area image processing model according to the physical separation area image and corresponding parameters and the current office physical security environment level;

[0057] The office environment image acquisition module is also used to receive the current physical separation area image and corresponding parameters, and is connected to the office area image processing module, and processes the current physical separation area image and corresponding parameters according to the office area image processing model in the office area image processing module to generate the current office physical security environment level;

[0058] In this embodiment, the office physical security environment level is two-dimensional data, one dimension of which can be divided into office physical security level and office confidentiality security level, and the other dimension is divided into low office physical security level, medium office physical security level, high office physical security level, low office confidentiality security level, medium office confidentiality security level, and high office confidentiality security level. The corresponding labels are 0, 1, and 2, which are the corresponding combinations of the above two security levels. For example, the label feature value corresponding to (low office physical security level, low office confidentiality security level) is (0,0), and the label feature value corresponding to (low office physical security level, medium office confidentiality security level) is (0,1). Others can be known by those skilled in the art with reference to the above, and will not be repeated here.

[0059] Since the model adopts a deep learning model, the training deep learning model uses the input physical separation area image feature information (T1, T2, ..., T1024, 26, 31, 1) and the corresponding label (0, 1) to obtain relevant model parameters. In this embodiment, a convolutional neural network can be used. The convolutional neural network is a conventional technical means for processing image data, which will not be repeated here. This embodiment can also use models such as BP neural network, which is not limited here.

[0060] The image acquisition module is also used to receive the current physical separation area image and corresponding parameters, and is connected to the office area image processing module, and processes the current physical separation area image and corresponding parameters according to the office area image processing model in the office area image processing module to generate the current office physical security environment level;

[0061] For example, the corresponding parameter characteristic value corresponding to the current physical separation area image is G(37,25,0), which is connected by the 1024-dimensional feature vector converted by the SM-ResNet50 model, and the physical separation area image feature information is (T1',T2',…,T1024',37,25,0). If the label result obtained by the office area image processing model at this time is (3,0), it means that this time it belongs to a high office physical security level and a low office confidentiality security level.

[0062] In this embodiment, the image features are processed to become high-dimensional data feature vectors.

[0063] The physical security environment alarm module receives the current office physical security environment level and issues an alarm level prompt based on the current office physical security environment level. The corresponding alarm levels are divided into three levels: low, medium and high. The medium level means that only one dimension has a medium security level, the high level means that two dimensions have a medium security level or one dimension has a high security level or above, and the others are low levels.

[0064] Furthermore, the office area image processing model is constructed using machine learning or deep learning methods.

[0065] Furthermore, the physical separation area image and corresponding parameters include:

[0066] The images of the physical separation area captured by the video acquisition device, the average of the parameters collected by the sensors of the physical separation area and the real-time parameters of the area given by the corresponding electronic equipment of the separation area, and the number of times the security alarm of the electronic equipment is triggered.

[0067] Furthermore, the improved regional image processing SM-ResNet50 model is used to process the physically separated regional images. The specific update rate σ t The formula is:

[0068]

[0069] Where: i is the change value; is the maximum value of the update rate, is the minimum value of the update rate, D cur is the current number of updates, D i is the total number of changes, The number of physical areas that the office area is physically separated into according to the number of times it has been updated.

[0070] Furthermore, the deep learning method uses an improved GAE function to train the model:

[0071]

[0072] In the formula, G(X) is the activation function value, is the number of physical areas divided by the physical separation area of ​​the office area according to the number of times it is updated, and x is the input feature value, which is obtained from the physical separation area image and corresponding parameters.

[0073] The physical security environment monitoring of the office area carried out by this method is safer and more accurate. On the one hand, it can improve the accuracy of the security level alarm of the office area monitoring. On the other hand, it integrates the image data of the office area and the parameters obtained by sensors and electronic devices online. It can not only obtain the specific disaster alarm level, but also monitor whether employees have issued an alarm for the leakage of commercial secrets and provide an alarm level. Finally, the acquired image data is also processed by the improved ResNet50 model, and the continuously changing image data is more suitable for processing in the specific application scenario of the office environment.

[0074] Of course, it can be understood that each embodiment of the present invention can achieve one of the effects alone, and a combination of multiple embodiments of the present invention can achieve all of the above effects, but it is not required that every embodiment of the present invention achieve all of the above advantages and effects, because each embodiment of the present invention can constitute a separate technical solution and make one or more contributions to the prior art.

[0075] The present invention does not particularly specify the structure of some modules, which shall be subject to the contents recorded in the prior art. The prior art mentioned in the above background technology section and the specific embodiment section of the present invention can be used as part of the present invention to understand the meaning of some technical features or parameters. The scope of protection of the present invention shall be subject to the contents actually recorded in the claims.

Claims

1. A method for monitoring the physical security environment of an office area, characterized by: Obtain the physical separation area image and corresponding parameters, and obtain the current office physical security environment level; Constructing an office area image processing model according to the physical separation area image and corresponding parameters and the current office physical security environment level; Acquire a current physical separation area image and corresponding parameters, process the current physical separation area image and corresponding parameters according to an office area image processing model, and generate a current office physical security environment level; Acquire the current office physical security environment level, and perform alarm processing according to the current office physical security environment level; The office area image processing model is constructed using a machine learning or deep learning method; Before constructing the office area image processing model according to the physical separation area image and corresponding parameters and the current office physical security environment level; and before processing the current physical separation area image and corresponding parameters according to the office area image processing model; The improved regional image processing SM-ResNet50 model is used to process the physical separation region image, and the specific update rate σ t The formula is: Where: i is the change value; is the maximum value of the update rate, is the minimum value of the update rate, D cur is the current number of updates, D i is the total number of changes, The number of physical areas divided by the physical partition area of ​​the office area according to the number of times it has been updated; The deep learning method uses an improved GAE function to train the model: In the formula, G(X) is the activation function value, is the number of physical areas divided by the physical separation area of ​​the office area according to the number of times it is updated, and x is the input feature value, which is obtained from the physical separation area image and corresponding parameters.

2. A method for monitoring the physical security environment of an office area as claimed in claim 1, characterized in that: Physical separation area images and corresponding parameters include: The images of the physical separation area captured by the video acquisition device, the average of the parameters collected by the sensors of the physical separation area and the real-time parameters of the area given by the corresponding electronic equipment of the separation area, and the number of times the security alarm of the electronic equipment is triggered.

3. A physical security environment monitoring system for an office area, characterized in that: The system comprises: An office environment image acquisition module, which acquires a physical separation area image and corresponding parameters, and acquires the current office physical security environment level according to the physical separation area image and corresponding parameters; An office area image processing module, which constructs an office area image processing model according to the physical separation area image and corresponding parameters and the current office physical security environment level; The office environment image acquisition module is also used to receive the current physical separation area image and corresponding parameters, and is connected to the office area image processing module, and processes the current physical separation area image and corresponding parameters according to the office area image processing model in the office area image processing module to generate the current office physical security environment level; A physical security environment alarm module receives the current office physical security environment level and issues an alarm level prompt according to the current office physical security environment level; The office area image processing model is constructed using a machine learning or deep learning method; Before constructing the office area image processing model according to the physical separation area image and corresponding parameters and the current office physical security environment level; and before processing the current physical separation area image and corresponding parameters according to the office area image processing model; The improved regional image processing SM-ResNet50 model is used to process the physical separation region image, and the specific update rate σ t The formula is: Where: i is the change value; is the maximum value of the update rate, is the minimum value of the update rate, D cur is the current number of updates, D i is the total number of changes, The number of physical areas divided by the physical partition area of ​​the office area according to the number of times it has been updated; The deep learning method uses an improved GAE function to train the model: In the formula, G(X) is the activation function value, is the number of physical areas divided by the physical separation area of ​​the office area according to the number of times it is updated, and x is the input feature value, which is obtained from the physical separation area image and corresponding parameters.

4. A physical security environment monitoring system for an office area as claimed in claim 3, characterized in that: Physical separation area images and corresponding parameters include: The images of the physical separation area captured by the video acquisition device, the average of the parameters collected by the sensors of the physical separation area and the real-time parameters of the area given by the corresponding electronic equipment of the separation area, and the number of times the security alarm of the electronic equipment is triggered.

Images