Security early warning system based on multi-video image analysis

Through the multi-video image analysis system, abnormal and dangerous behavior coefficients are calculated, and the problem of failure to effectively utilize the stay time and behavior weight of unauthorized personnel in the existing technology is solved, real-time early warning and security assessment of the community monitoring area is realized, and early warning is improved Responsiveness.

CN119007082AInactive Publication Date: 2024-11-22ANHUI POLYTECHNIC UNIV MECHANICAL & ELECTRICAL COLLEGE
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202411125446.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-16
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing security warning system is difficult to effectively utilize the proportional data of unauthorized personnel's stay time, and cannot integrate and analyze the weights of each behavior, resulting in the inability to accurately calculate the coefficients, lack of comprehensive evaluation of each behavior coefficient based on calculations, making it difficult to early warning of potential hazard information, resulting in The early warning measures are not responded in time.

Method used

Using a multi-video image analysis system, including image acquisition, image processing, image analysis and early warning modules, analyze safety hazards through face comparison and deep learning models, calculate abnormal behavior coefficients and dangerous behavior coefficients, comprehensively evaluate the impact safety coefficients, and generate early warnings. The information performs early warning operations.

Benefits of technology

Real-time early warning of community monitoring areas is achieved, the response speed and accuracy of potential hazard information is improved, and the safety of residents' lives and property is ensured.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119007082A_ABST
    Figure CN119007082A_ABST
Patent Text Reader

Abstract

The invention discloses a security and protection early warning system based on multi-video image analysis, relates to the field of security and protection early warning, and solves the problems that an existing security and protection early warning system is difficult to effectively utilize proportion data of stay time of unauthorized personnel and cannot integrate and analyze each behavior weight, so that a coefficient cannot be accurately calculated; and meanwhile, a technical problem that early warning measures are not timely in response because a comprehensive evaluation cell influence safety coefficient based on each calculated behavior coefficient is lacked and early warning of potential danger information is difficult to carry out is solved. The system comprises the following modules: an image acquisition module, an image processing module, an image analysis module and an early warning module. The abnormal behavior coefficient and the dangerous behavior coefficient are calculated through various data. And comprehensively evaluating a cell influence safety coefficient based on each calculated behavior coefficient, thereby accurately generating early warning information to realize real-time early warning of a cell monitoring area.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of security warning, and in particular to a security warning system based on multi-video image analysis. Background Art

[0002] With the progress of the times and the innovation of science and technology, the security early warning system can display the data of monitoring equipment in a centralized manner to realize real-time monitoring. The present invention determines the category of people entering the community through face comparison. By analyzing the potential safety hazards and limiting the stay time of unauthorized personnel, the abnormal behavior coefficient and the dangerous behavior coefficient are calculated and the safety factor is comprehensively evaluated. According to the safety factor, early warning information is generated and early warning operations are performed. Real-time early warning of the community monitoring area is realized, the life and property safety of community residents is guaranteed, and a safe living environment is provided for community residents.

[0003] The existing security warning system cannot effectively utilize the proportion of unauthorized personnel's stay time, and cannot integrate and analyze the weights of various behaviors, resulting in the inability to accurately calculate their coefficients. At the same time, there is a lack of comprehensive assessment of the safety factor of the community based on the calculated coefficients of various behaviors, making it difficult to issue warnings for potential danger information, resulting in untimely response to warning measures. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a security warning system based on multi-video image analysis, which is used to solve the problem that the existing security warning system is difficult to effectively utilize the proportion data of the stay time of unauthorized personnel, and is unable to integrate and analyze the weights of various behaviors, resulting in the inability to accurately calculate their coefficients; at the same time, there is a lack of a comprehensive assessment of the safety factor of the impact of each behavior coefficient based on the calculation, making it difficult to warn of potential danger information, resulting in the technical problem that the warning measures are not responded to in a timely manner.

[0005] To solve the above problems, the first aspect of the present invention provides a security warning system based on multi-video image analysis, comprising the following modules:

[0006] Image acquisition module: used to acquire video images of the monitoring area in real time;

[0007] Image processing module: used to pre-process the collected video images, including: denoising and image enhancement;

[0008] Image analysis module: used to compare faces on pre-processed video images to determine the types of people entering the community; analyze potential safety hazards in the community and limit the length of stay of unauthorized people entering the community; obtain abnormal behavior coefficients and dangerous behavior coefficients based on the proportion of unauthorized people's length of stay and the behavioral weights of abnormal and dangerous behaviors;

[0009] Early warning module: Based on the abnormal behavior coefficient and dangerous behavior coefficient of the community, comprehensively evaluate the impact safety factor of the community, and generate early warning information and perform early warning operations based on the impact safety factor of the community.

[0010] As a further solution of the present invention, face comparison is performed on the pre-processed video image to determine the category of the person entering the community, including the following steps:

[0011] Performing face comparison detection on the preprocessed video image to determine the face area in the preprocessed video image;

[0012] Extracting feature points of the face area;

[0013] Performing feature description on the feature points of the face region to obtain a feature vector of the face region;

[0014] According to the feature vector of the face area, a product quantization method is used to search for feature vectors of community residents in the feature vector of the face area for comparison, so as to obtain a comparison result corresponding to the face area;

[0015] Based on the face comparison results, the categories of people entering the community are determined to include residents and unauthorized persons. The access request records are queried to determine that the categories of unauthorized persons include visitors and non-visitors.

[0016] As a further solution of the present invention: analyzing the potential safety hazards in the community includes the following steps:

[0017] Based on the pre-processed video images, the deep learning model is used to analyze the potential safety hazards of the community in the video images;

[0018] Mark the potential safety hazards in the community, where potential safety hazards include n types of abnormal behaviors and m types of dangerous behaviors;

[0019] Input the labeled i-th abnormal behavior and j-th dangerous behavior into the convolutional neural network of the deep learning model for autonomous learning and training, where i∈(1,2,…,n) and j∈(1,2,…,m);

[0020] The security early warning system collects and processes video images of the monitored area in real time to identify the i-th abnormal behavior and the j-th dangerous behavior in the community.

[0021] As a further solution of the present invention: n kinds of abnormal behaviors and m kinds of dangerous behaviors include:

[0022] The anomalies marked include: the first type is the presence of unauthorized personnel, the second type is the random parking of vehicles, and the third type is littering;

[0023] The marked dangerous behaviors include: the first type is that unauthorized persons stay for longer than the maximum permitted stay time, the second type is that unauthorized persons climb over walls or fences, the third type is that they climb over anti-theft windows, the fourth type is that physical conflicts occur, and the fifth type is that objects are thrown from high places.

[0024] As a further solution of the present invention: limiting the stay time of unauthorized persons entering the community, including the following steps:

[0025] S1: Basic stay time: The basic time that unauthorized persons entering the community are allowed to stay in the community without additional permission. The basic time set by the property management is Tbase;

[0026] S2: Extra time: In special circumstances, unauthorized persons entering the community visit specific residents or conduct business activities. Visitors apply for additional stay time, which is approved by the property management. The visitor application information is entered into the access request record. The additional application approval time is T extra;

[0027] S3: Total stay time: the total stay time of visitors entering the community, the basic stay time of unauthorized persons entering the community and the additional stay time of visitors applying for approval, Ttotal = Tbasic+Textra;

[0028] S4: Maximum allowed stay time: The total stay time of unauthorized persons entering the community should not exceed the maximum allowed stay time, which is set by the property management as Tmax.

[0029] As a further solution of the present invention, the abnormal behavior coefficient and the dangerous behavior coefficient are obtained by the proportion of the unauthorized personnel's stay time and the behavior weights of the abnormal behavior and the dangerous behavior, including the following steps:

[0030] By analyzing the formula:

[0031]

[0032] The abnormal behavior coefficient α of the community is obtained, where Ttotal represents the total stay time of visitors entering the community, Tmax represents the maximum allowable stay time of unauthorized persons entering the community, Tactual represents the actual stay time of non-visitors, and w i is the weight of the ith abnormal behavior, x i The indicators marked as abnormal behaviors include the number of vehicles parked randomly and the number of times littering;

[0033] By analyzing the formula:

[0034] The dangerous behavior coefficient β of the community is obtained, where Tunauthorized represents the actual stay time of unauthorized persons entering the community that exceeds the maximum allowable stay time, Tmax represents the maximum allowable stay time of unauthorized persons entering the community, and v j is the weight of the j-th dangerous behavior, y j The indicators of the jth dangerous behavior include the number of times unauthorized persons climb over walls or fences, the height of climbing anti-theft windows, the number of people involved in physical conflicts, and the number of floors from which objects are thrown from high places.

[0035] As a further solution of the present invention: based on the abnormal behavior coefficient and the dangerous behavior coefficient of the cell, comprehensively evaluate the impact safety coefficient of the cell, and generate warning information according to the impact safety coefficient of the cell to perform a warning operation, including the following steps:

[0036] By analyzing the formula:

[0037] The impact safety factor Q of the cell is obtained, where α represents the abnormal behavior coefficient of the cell and β represents the dangerous behavior coefficient of the cell.

[0038] As a further solution of the present invention: generating early warning information according to the impact safety factor of the cell and performing early warning operation, including the following steps:

[0039] According to the value of the safety factor Q, different levels of thresholds are set, and the warning is divided into three levels: first-level warning, second-level warning, and third-level warning;

[0040] When Q<1, the warning is divided into level 1 warning and a warning signal is issued;

[0041] When 1≤Q<1.5, the warning is divided into two levels, and a warning signal is issued to remind security personnel to check the monitoring;

[0042] When Q ≥ 1.5, the warning is divided into three levels, a serious warning signal is issued, and security personnel are summoned for security inspection.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] The present invention solves the technical problems that the existing security warning system is difficult to effectively utilize the proportion data of the unauthorized personnel's stay time, and is unable to integrate and analyze the weights of various behaviors, resulting in the inability to accurately calculate their coefficients; at the same time, there is a lack of comprehensive evaluation of the community impact safety coefficient based on the calculated coefficients of various behaviors, making it difficult to issue warnings for potential dangerous information, resulting in untimely response to warning measures. The present invention includes the following modules: image acquisition module, image processing module, image analysis module and warning module. It realizes the calculation of the impact safety coefficient based on the abnormal behavior coefficient and the dangerous behavior coefficient, and generates warning information based on the impact safety coefficient to achieve real-time warning of the community monitoring area. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 This is a schematic diagram of the system module framework of the present invention;

[0047] Figure 2 is a flow chart of the method of the present invention; DETAILED DESCRIPTION

[0048] The technical scheme of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] See also Figure 1 The first aspect of the present invention provides a security warning system based on multi-video image analysis, comprising the following modules:

[0050] Image acquisition module: used to acquire video images of the monitoring area in real time;

[0051] Image processing module: used to pre-process the collected video images, including: denoising and image enhancement;

[0052] Image analysis module: used to compare faces on pre-processed video images to determine the types of people entering the community; analyze potential safety hazards in the community and limit the length of stay of unauthorized people entering the community; obtain abnormal behavior coefficients and dangerous behavior coefficients based on the proportion of unauthorized people's length of stay and the behavioral weights of abnormal and dangerous behaviors;

[0053] Early warning module: Based on the abnormal behavior coefficient and dangerous behavior coefficient of the community, comprehensively evaluate the impact safety factor of the community, and generate early warning information and perform early warning operations based on the impact safety factor of the community.

[0054] Specifically, the present invention collects video images of the monitoring area in real time through an image acquisition module, an image processing module, an image analysis module and an early warning module for preprocessing, removes noise from the collected video images, adjusts and enhances the contrast, brightness, smoothness and edge information of the video images to make the details of the video images clearer, performs face comparison on the preprocessed video images to determine the category of people entering the community; analyzes the safety hazards in the community and limits the stay time of unauthorized people entering the community; obtains the abnormal behavior coefficient and the dangerous behavior coefficient through the proportion of the stay time of unauthorized people and the behavioral weights of abnormal behaviors and dangerous behaviors; comprehensively evaluates the impact safety factor of the community based on the abnormal behavior coefficient and the dangerous behavior coefficient of the community, and generates early warning information based on the impact safety factor to execute early warning operations.

[0055] See also Figure 2 In one embodiment of the present invention, face comparison is performed on the pre-processed video image to determine the type of person entering the community, including the following steps:

[0056] Performing face comparison detection on the preprocessed video image to determine the face area in the preprocessed video image;

[0057] Extracting feature points of the face area;

[0058] Performing feature description on the feature points of the face region to obtain a feature vector of the face region;

[0059] According to the feature vector of the face area, a product quantization method is used to search for feature vectors of community residents in the feature vector of the face area for comparison, so as to obtain a comparison result corresponding to the face area;

[0060] Based on the face comparison results, the categories of people entering the community are determined to include residents and unauthorized persons. The access request records are queried to determine that the categories of unauthorized persons include visitors and non-visitors.

[0061] Specifically, face comparison detection is performed on the preprocessed video image to determine the face image area entering the community in the preprocessed video image; feature points of the face area are extracted according to an algorithm; feature description is performed on the feature points of the face area to obtain a feature vector of the face area; the extracted high-dimensional feature vector is quantized using a product quantization method and converted into a low-bit rate vector, and the Hamming distance is calculated between the quantized features, that is, the number of different corresponding bits between two binary sequences, so as to evaluate the similarity between faces and obtain a comparison result; according to the face comparison result, the one with high similarity is a resident in the community, otherwise it is an unauthorized person, and the access request record is queried to determine whether the unauthorized person is a visitor or a non-visitor.

[0062] In one embodiment of the present invention, analyzing the potential safety hazard in a cell includes the following steps:

[0063] Based on the pre-processed video images, the deep learning model is used to analyze the potential safety hazards of the community in the video images;

[0064] Mark the potential safety hazards in the community, where potential safety hazards include n types of abnormal behaviors and m types of dangerous behaviors;

[0065] Input the labeled i-th abnormal behavior and j-th dangerous behavior into the convolutional neural network of the deep learning model for autonomous learning and training, where i∈(1,2,…,n) and j∈(1,2,…,m);

[0066] The security early warning system collects and processes video images of the monitored area in real time to identify the i-th abnormal behavior and the j-th dangerous behavior in the community.

[0067] Specifically, based on the preprocessed video images, a deep learning model is constructed to analyze the potential safety hazards of the community in the video images. The deep learning model can analyze a large amount of data and is usually composed of multiple hidden layers, each of which contains a large number of neurons. This multi-layer structure enables the model to learn the deep-level features and complex patterns of the data. Carefully annotate various potential safety hazards presented in the video images, where the annotated safety hazards include n abnormal behaviors and m dangerous behaviors; input the annotated i-th abnormal behavior and j-th dangerous behavior into the convolutional neural network of the deep learning model for autonomous learning and training, where i∈(1,2,…,n), j∈(1,2,…,m); collect and process the video images of the monitored area in real time through the security early warning system, and identify the i-th abnormal behavior and j-th dangerous behavior in the community.

[0068] In one embodiment of the present invention, the n abnormal behaviors and the m dangerous behaviors include:

[0069] The anomalies marked include: the first type is the presence of unauthorized personnel, the second type is the random parking of vehicles, and the third type is littering;

[0070] The marked dangerous behaviors include: the first type is that unauthorized persons stay for longer than the maximum permitted stay time, the second type is that unauthorized persons climb over walls or fences, the third type is that they climb over anti-theft windows, the fourth type is that physical conflicts occur, and the fifth type is that objects are thrown from high places.

[0071] Specifically, the marked abnormal behaviors include: the first type of unauthorized persons appearing in the video images, the second type of vehicles being parked randomly in the video surveillance area of ​​the community, and the third type of people littering in the community; the marked dangerous behaviors include: the first type of unauthorized persons in the video images staying longer than the maximum allowable stay time set by the property management, the second type of unauthorized persons climbing over the walls or fences of the community, the third type of people in the community climbing the anti-theft windows of residential buildings, the fourth type of physical conflicts between people in the community in the video images, and the fifth type of people in the community throwing objects from high places in residential buildings.

[0072] In one embodiment of the present invention, limiting the stay time of unauthorized persons entering a community includes the following steps:

[0073] S1: Basic stay time: The basic time that unauthorized persons entering the community are allowed to stay in the community without additional permission. The basic time set by the property management is Tbase;

[0074] S2: Extra time: In special circumstances, unauthorized persons entering the community visit specific residents or conduct business activities. Visitors apply for additional stay time, which is approved by the property management. The visitor application information is entered into the access request record. The additional application approval time is T extra;

[0075] S3: Total stay time: the total stay time of visitors entering the community, the basic stay time of unauthorized persons entering the community and the additional stay time of visitors applying for approval, Ttotal = Tbasic+Textra;

[0076] S4: Maximum allowed stay time: The total stay time of unauthorized persons entering the community should not exceed the maximum allowed stay time, which is set by the property management as Tmax.

[0077] Specifically, in order to ensure the safety and order of the community, it is usually necessary to limit the stay time of unauthorized persons entering the community. This time can be set according to the actual situation and management policy of the community, and may be adjusted under special circumstances. The property management sets the basic stay time for unauthorized persons based on the actual situation of the community. Under special circumstances, unauthorized persons who need to visit specific residents or conduct business activities must apply to the property management for additional stay time. After approval, the information will be entered into the access request record. The total stay time of unauthorized persons entering the community is the sum of their basic stay time and the time for additional application approval. The above total stay time should not exceed the maximum allowable stay time set by the property management. If the total stay time of unauthorized persons exceeds the maximum allowable stay time set by the property management, there will be safety hazards.

[0078] In one embodiment of the present invention, the abnormal behavior coefficient and the dangerous behavior coefficient are obtained by using the proportion of the unauthorized personnel's stay time and the behavior weights of the abnormal behavior and the dangerous behavior, including the following steps:

[0079] By analyzing the formula:

[0080]

[0081] The abnormal behavior coefficient α of the community is obtained, where Ttotal represents the total stay time of visitors entering the community, Tmax represents the maximum allowable stay time of unauthorized persons entering the community, Tactual represents the actual stay time of non-visitors, and w i is the weight of the ith abnormal behavior, x i The indicators marked as abnormal behaviors include the number of vehicles parked randomly and the number of times littering;

[0082] By analyzing the formula:

[0083] The dangerous behavior coefficient β of the community is obtained, where Tunauthorized represents the actual stay time of unauthorized persons entering the community that exceeds the maximum allowable stay time, Tmax represents the maximum allowable stay time of unauthorized persons entering the community, and v j is the weight of the j-th dangerous behavior, y j The indicators of the jth dangerous behavior include the number of times unauthorized persons climb over walls or fences, the height of climbing anti-theft windows, the number of people involved in physical conflicts, and the number of floors from which objects are thrown from high places.

[0084] Specifically, the abnormal behavior coefficient and the dangerous behavior coefficient are calculated by the proportion of the unauthorized personnel's stay time and the behavioral weights of abnormal behaviors and dangerous behaviors. First, the abnormal behavior coefficient can be calculated based on the proportional relationship between the unauthorized personnel's stay time in the community and the maximum allowable stay time. In addition, other detected abnormal behaviors should also be taken into consideration, and the n types of abnormal behaviors are given different weights according to the severity of their nature. Secondly, the dangerous behavior coefficient is calculated based on the proportional relationship between the actual stay time of unauthorized personnel entering the community that exceeds the maximum allowable stay time and the maximum allowable stay time. This is intended to capture those potential dangerous behaviors that pose a direct threat to community safety. Similarly, the m types of dangerous behaviors detected should be assigned different weights according to their direct threat level to community safety.

[0085] In one embodiment of the present invention, based on the abnormal behavior coefficient and the dangerous behavior coefficient of the cell, the safety factor of the cell is comprehensively evaluated, and according to the safety factor of the cell, warning information is generated to perform a warning operation, including the following steps:

[0086] By analyzing the formula:

[0087] The impact safety factor Q of the cell is obtained, where α represents the abnormal behavior coefficient of the cell and β represents the dangerous behavior coefficient of the cell.

[0088] Specifically, the impact safety factor of the community is comprehensively evaluated based on the abnormal behavior coefficient and dangerous behavior coefficient of the community. The impact safety factor Q is obtained by calculating the ratio of the product of the abnormal behavior coefficient α and the dangerous behavior coefficient β to the square root of the sum of the abnormal behavior coefficient α and the dangerous behavior coefficient β plus a constant. The value of the impact safety factor Q can be used to judge the safety hazards of the security early warning system, so as to achieve real-time monitoring and early warning.

[0089] In one embodiment of the present invention, generating warning information according to the impact safety factor of the cell and performing a warning operation include the following steps:

[0090] According to the value of the safety factor Q, different levels of thresholds are set, and the warning is divided into three levels: first-level warning, second-level warning, and third-level warning;

[0091] When Q<1, the warning is divided into level 1 warning and a warning signal is issued;

[0092] When 1≤Q<1.5, the warning is divided into two levels, and a warning signal is issued to remind security personnel to check the monitoring;

[0093] When Q ≥ 1.5, the warning is divided into three levels, a serious warning signal is issued, and security personnel are summoned for security inspection.

[0094] Specifically, different levels of thresholds are set according to the value of the influencing safety factor Q, and the warning is divided into three levels: primary warning, secondary warning, and tertiary warning. The value of the influencing safety factor Q calculated in real time is compared with the threshold to generate warning information and execute corresponding warning operations.

[0095] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A security warning system based on multi-video image analysis, characterized in that: Includes the following modules: Image acquisition module: used to acquire video images of the monitoring area in real time; Image processing module: used to pre-process the collected video images, including: denoising and image enhancement; Image analysis module: used to compare faces on pre-processed video images to determine the types of people entering the community; analyze potential safety hazards in the community and limit the length of stay of unauthorized people entering the community; obtain abnormal behavior coefficients and dangerous behavior coefficients based on the proportion of unauthorized people's length of stay and the behavioral weights of abnormal and dangerous behaviors; Early warning module: Based on the abnormal behavior coefficient and dangerous behavior coefficient of the community, comprehensively evaluate the impact safety factor of the community, and generate early warning information and perform early warning operations based on the impact safety factor of the community.

2. A security warning system based on multi-video image analysis according to claim 1, characterized in that: Performing face comparison on the pre-processed video image to determine the type of person entering the community includes the following steps: Performing face comparison detection on the preprocessed video image to determine the face area in the preprocessed video image; Extracting feature points of the face area; Performing feature description on the feature points of the face region to obtain a feature vector of the face region; According to the feature vector of the face area, a product quantization method is used to search for feature vectors of community residents in the feature vector of the face area for comparison, so as to obtain a comparison result corresponding to the face area; Based on the face comparison results, the categories of people entering the community are determined to include residents and unauthorized persons. The access request records are queried to determine that the categories of unauthorized persons include visitors and non-visitors.

3. The security warning system based on multi-video image analysis according to claim 1 is characterized in that: Analyze the potential safety hazards in the community, including the following steps: Based on the pre-processed video images, the deep learning model is used to analyze the potential safety hazards of the community in the video images; Mark the potential safety hazards in the community, where potential safety hazards include n types of abnormal behaviors and m types of dangerous behaviors; Input the labeled i-th abnormal behavior and j-th dangerous behavior into the convolutional neural network of the deep learning model for autonomous learning and training, where i∈(1,2,…,n) and j∈(1,2,…,m); The security early warning system collects and processes video images of the monitored area in real time to identify the i-th abnormal behavior and the j-th dangerous behavior in the community.

4. The security warning system based on multi-video image analysis according to claim 3 is characterized in that: The n abnormal behaviors and m dangerous behaviors include: The anomalies marked include: the first type is the presence of unauthorized personnel, the second type is the random parking of vehicles, and the third type is littering; The marked dangerous behaviors include: the first type is that unauthorized persons stay for longer than the maximum permitted stay time, the second type is that unauthorized persons climb over walls or fences, the third type is that they climb over anti-theft windows, the fourth type is that physical conflicts occur, and the fifth type is that objects are thrown from high places.

5. The security warning system based on multi-video image analysis according to claim 1 is characterized in that: Limit the time that unauthorized persons can stay in the community, including the following steps: S1: Basic stay time: The basic time that unauthorized persons entering the community are allowed to stay in the community without additional permission. The basic time set by the property management is Tbase; S2: Extra time: In special circumstances, unauthorized persons entering the community visit specific residents or conduct business activities. Visitors apply for additional stay time, which is approved by the property management. The visitor application information is entered into the access request record. The additional application approval time is T extra; S3: Total stay time: the total stay time of visitors entering the community, the basic stay time of unauthorized persons entering the community and the additional stay time of visitors applying for approval, Ttotal = Tbasic+Textra; S4: Maximum allowed stay time: The total stay time of unauthorized persons entering the community should not exceed the maximum allowed stay time, which is set by the property management as Tmax.

6. The security warning system based on multi-video image analysis according to claim 1, characterized in that: The abnormal behavior coefficient and the dangerous behavior coefficient are obtained by the proportion of the unauthorized personnel's stay time and the behavior weights of abnormal behavior and dangerous behavior, including the following steps: By analyzing the formula: The abnormal behavior coefficient α of the community is obtained, where Ttotal represents the total stay time of visitors entering the community, Tmax represents the maximum allowable stay time of unauthorized persons entering the community, Tactual represents the actual stay time of non-visitors, and w i is the weight of the ith abnormal behavior, x i The indicators marked as abnormal behaviors include the number of vehicles parked randomly and the number of times littering; By analyzing the formula: The dangerous behavior coefficient β of the community is obtained, where Tunauthorized represents the actual stay time of unauthorized persons entering the community that exceeds the maximum allowable stay time, Tmax represents the maximum allowable stay time of unauthorized persons entering the community, and v j is the weight of the j-th dangerous behavior, y j The indicators of the jth dangerous behavior include the number of times unauthorized persons climb over walls or fences, the height of climbing anti-theft windows, the number of people involved in physical conflicts, and the number of floors from which objects are thrown from high places.

7. The security warning system based on multi-video image analysis according to claim 1, characterized in that: According to the abnormal behavior coefficient and dangerous behavior coefficient of the cell, the safety factor of the cell is comprehensively evaluated, and warning information is generated according to the safety factor of the cell to perform warning operations, including the following steps: By analyzing the formula: The impact safety factor Q of the cell is obtained, where α represents the abnormal behavior coefficient of the cell and β represents the dangerous behavior coefficient of the cell.

8. The security warning system based on multi-video image analysis according to claim 7, characterized in that: According to the safety factor of the cell, the warning information is generated and the warning operation is performed, including the following steps: According to the value of the safety factor Q, different levels of thresholds are set, and the warning is divided into three levels: first-level warning, second-level warning, and third-level warning; When Q<1, the warning is divided into level 1 warning and a warning signal is issued; When 1≤Q<1.5, the warning is divided into two levels, and a warning signal is issued to remind security personnel to check the monitoring; When Q ≥ 1.5, the warning is divided into three levels, a serious warning signal is issued, and security personnel are summoned for security inspection.

Citation Information

Cited By

  • Multi-level analysis and monitoring system capable of integrating multiple video streams in real time

    CN119963823A

  • A real-time integrated multi-video stream multi-level analysis and monitoring system

    CN119963823B

  • Intelligent security and protection monitoring method and system based on AI behavior analysis and medium

    CN120086664A