Campus safety management method and system

Abstract

The invention relates to the technical field of public opinion analysis, in particular to a campus security management method and system. According to the technical scheme, a campus sensitive area is determined; acquiring image information and / or voice information of a person entering the sensitive area in real time, and analyzing whether the person in the image has an abnormal behavior or not based on the image information and / or the voice information; if yes, alarm operation is executed, and the object is marked as a key concerned object; acquiring feature information of school entering personnel; judging whether the target personnel belong to on-campus personnel or off-campus personnel based on the feature information; if so, adopting a first management method; if yes, judging whether the target person belongs to a key concerned object or not; if yes, adopting a second management method; otherwise, adopting a third management method. By adopting the scheme provided by the invention, the probability of occurrence of unsafe events in a campus can be reduced, and the efficiency and accuracy of abnormal behavior detection are improved.

Description

Technical Field

[0001] The present application relates to the technical field of public opinion analysis, and specifically to a campus safety management method and system. Background Art

[0002] In the prior art, although cameras are widely used to monitor various areas of the school, monitoring personnel are required to monitor the images of multiple cameras in real time. Although this monitoring method improves the safety of the campus to some extent, due to the large number of areas and people on the campus, it is often difficult for monitoring personnel to detect abnormal behavior and stop it in time. Therefore, this traditional monitoring method has certain limitations in dealing with campus insecurity issues. Summary of the invention

[0004] In order to overcome the above problems or partially solve the above problems, the purpose of this application is to provide a campus safety management method and system to reduce the probability of unsafe incidents on campus and improve the efficiency and accuracy of abnormal behavior detection.

[0005] This application is implemented through the following technical solutions:

[0006] In a first aspect, the present application provides a campus safety management method, the method comprising: determining sensitive areas on campus; obtaining image information and / or voice information of people entering the sensitive areas in real time, and analyzing whether the people in the image have abnormal behavior based on the image information and / or voice information; the abnormal behavior includes violent behavior and self-harm behavior; if so, executing an alarm operation, and at the same time obtaining the identity information of the person with abnormal behavior based on the image information and / or voice information, and marking it as a key focus object; obtaining feature information of people entering the school; judging whether the target person belongs to an on-campus person or an off-campus person based on the feature information; if the target person is an off-campus person, adopting the first management method; if the target person is an on-campus person, judging whether the target person is a key focus object; if so, adopting the second management method; otherwise adopting the third management method; the second management method comprises: calculating the correlation between any two key focus objects, and if the correlation is greater than a preset correlation threshold, the two key focus objects are grouped or managed in classes.

[0007] Based on the first aspect, in an embodiment of the present application, determining sensitive areas on campus includes: counting areas where historical abnormal behaviors occurred on campus; determining areas where historical abnormal behaviors occurred A times or more as sensitive areas on campus, and the remaining areas as non-sensitive areas; wherein A is a set value.

[0008] Based on the first aspect, in an embodiment of the present application, the first management method includes: obtaining and recording facial feature information, time information of entering and leaving the campus, behavioral information and movement trajectory of non-school personnel; judging whether the non-school personnel are suspicious based on the behavioral information and movement trajectory within the campus; if so, alarming and intervening; obtaining unexpected events that occur on campus within a preset time period after the non-school personnel leave the school, and analyzing the correlation between the non-school personnel and the unexpected events based on the time information of entering and leaving the campus, behavioral information and movement trajectory within the campus.

[0009] Based on the first aspect, in an embodiment of the present application, if the identity of the on-campus personnel is a student, the second management method or the third management method includes: obtaining the time information of the students entering and leaving the campus, and synchronizing the time information to the mobile terminal of the student's guardian in real time.

[0010] Based on the first aspect, in an embodiment of the present application, the second management method includes: in a non-sensitive area of ​​the campus, obtaining the behavior information of the target person at a first frequency, analyzing whether the target person has abnormal behavior based on the behavior information, and if so, alarming and intervening; the third management method includes: in a non-sensitive area of ​​the campus, obtaining the behavior information of the target person at a second frequency, analyzing whether the target person has abnormal behavior based on the behavior information, and if so, alarming and intervening; recording the number of abnormal behaviors of the target person, and if the number of abnormal behaviors exceeds a threshold, marking the target person as a key focus object; wherein the first frequency is higher than the second frequency.

[0011] Based on the first aspect, in the embodiment of the present application, in the second management method, the steps of calculating the correlation G between two focus objects a and b are as follows: obtaining an image video in which the focus object a or b exists in the screen other than the classroom on campus; randomly extracting N frames of images from the image video, and counting the number of image frames N in which the focus objects a and b exist at the same time in the N frames of images. ab , count the number of image frames N0 belonging to class collective activities in N frames of images; correlation G = (N ab -N0) / N(1).

[0012] In a second aspect, the present application provides a campus security management system, the system comprising: a division module for dividing the campus into sensitive areas and non-sensitive areas; a real-time monitoring module, arranged in the sensitive area, for acquiring image information and / or voice information of people entering the sensitive area in real time, and analyzing whether there is abnormal behavior of the people in the image based on the image information and / or voice information; the abnormal behavior includes violent behavior and self-harm behavior; an alarm module, for executing an alarm operation when abnormal behavior occurs; a marking module, for acquiring the identity information of the person with abnormal behavior based on the image information and / or voice information, and marking it as a key focus object; a feature acquisition module, for acquiring feature information of people entering the school; a feature discrimination module, for judging whether the target person belongs to an on-campus person or an off-campus person based on the feature information; a management control module, for managing off-campus people using a first management method, managing key focus objects using a second management method, and managing non-key focus objects using a third management method; the second management method comprises: calculating the correlation between any two key focus objects, and if the correlation is greater than a preset correlation threshold, the two key focus objects are grouped or managed in classes.

[0013] Based on the second aspect, in an embodiment of the present application, the system also includes: a fixed monitoring component, which is arranged in a non-sensitive area, and is used to obtain image information of people in the non-sensitive area, and extract facial feature information and behavior information of the target person based on the image information; a mobile monitoring component, which is used to patrol monitoring blind spots in sensitive areas and monitoring blind spots in non-sensitive areas, and receive commands from the management control module to go to a specific location to perform monitoring tasks.

[0014] Based on the second aspect, in an embodiment of the present application, the mobile monitoring component includes: a robot or a drone, on which an intervention module is installed, and the intervention module is used to intervene in the abnormal behavior of the target person; if the target person has abnormal behavior, the management control module is also used to obtain the location of the target person, and drive the robot or drone to the location of the target person to intervene in the abnormal behavior of the target person.

[0015] Based on the second aspect, in the embodiment of the present application, the robot or drone includes two working modes: daily inspection mode and target tracking mode; the daily inspection mode is: to order the robot or drone to regularly perform monitoring tasks according to a pre-planned fixed inspection route; the target tracking mode is: when abnormal behavior is found, track the abnormal person and record the abnormal person's route and image information until a command to stop tracking is received or the abnormal person leaves the monitoring range of the robot or drone.

[0016] Based on the second aspect, in an embodiment of the present application, the sensitive area includes a bathroom, and the real-time monitoring module provided in the bathroom includes a voice detection unit, which is used to obtain voice information in the bathroom and analyze whether there is abnormal behavior in the bathroom based on the voice information.

[0017] Based on the second aspect, in an embodiment of the present application, the target tracking mode execution steps of the drone include:

[0018] B1. Obtain the coordinates (x0, y0, z0) of the person with abnormal behavior in a pre-established three-dimensional coordinate system;

[0019] B2. Establish the three-dimensional kinematic model of the UAV: In formula (2), x, y, z are the position coordinates of the UAV, v is the speed of the UAV, α is the heading angle of the UAV, β is the track angle of the UAV, and ω is the control variable;

[0020] B3. Set the tracking and guidance law of the UAV: In the above formula (3), θ represents the angle between the speed of the drone and the relative distance, and the relative distance is the relative distance between the drone and the abnormal person, where: λ represents the straight-line distance between the drone and the abnormal actor, and its expression is: L represents the expected distance between the UAV and the abnormal actor, which is set to be known; h represents the height of the UAV, where h = z = vsinβ (6); H represents the expected height value of the UAV, and H = h0 + ΔH (7), where h0 represents the height of the abnormal actor, and ΔH represents the fixed height difference between the abnormal actor and the UAV; μ1 and μ2 are adjustment parameters;

[0021] B4. Calculate the control quantity according to the three-dimensional kinematic model, the tracking guidance law of the UAV and equations (4), (5), (6) and (7); the control quantity includes the desired heading angle of the UAV;

[0022] B5. Update the control quantity and control the drone to track the person with abnormal behavior.

[0023] Compared with the prior art, this application has at least the following advantages and beneficial effects:

[0024] The solution provided by this application divides the people entering the school into three categories according to their identities and behaviors, and manages them with different levels of attention. Special attention is paid to people outside the school and sensitive personnel, while other normal personnel can reduce the computing burden of the server by reducing the level of attention, thereby improving the efficiency and accuracy of identifying abnormal behaviors. In addition, this application also expands the monitoring range by adding a mobile monitoring component, and uses the mobile monitoring component to regularly patrol the monitoring blind spots of the fixed monitoring module, which can greatly reduce the missed detection rate of abnormal behaviors. And because the mobile monitoring component can track and shoot close to the target, the accuracy of abnormal behavior detection is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present application and should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:

[0026] Figure 1 A schematic flow chart of the campus safety management method in the embodiment;

[0027] Figure 2 It is a schematic block diagram of the structure of the campus security management system in the embodiment.

[0028] icon:

[0029] 1-division module; 2-real-time monitoring module; 3-alarm module; 4-marking module; 5-feature acquisition module; 6-feature identification module; 7-management and control module; 8-fixed monitoring component; 9-mobile monitoring component. DETAILED DESCRIPTION

[0030] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with examples and drawings. The illustrative implementation scheme of the present application and its description are only used to explain the present application and are not intended to limit the present application. It should be noted that the audio data, video data, etc. used in the embodiments of the present invention are collected and used with the consent of the school.

[0031] Example 1

[0032] This embodiment provides a campus safety management method, please refer to Figure 1 , the method comprising:

[0033] S1. Identify sensitive areas on campus;

[0034] The specific determination methods include:

[0035] Statistics are collected on the areas where historical abnormal behaviors occurred on campus; areas where historical abnormal behaviors occurred A times or more are identified as sensitive areas on campus, and the rest are non-sensitive areas; A is a set value; the A value can be set to 0 or 1.

[0036] In addition to collecting historical data from our own school, we can also refer to historical data from schools around the world. The areas with high incidence of abnormal behavior are usually the top floor, bathroom, student dormitory, etc., which are rarely visited by teachers and leaders. Of course, the architectural layout of each school is different, so there may be differences. In addition to the above areas, it may also include libraries, equipment rooms, etc., which can be determined according to the specific situation of the school.

[0037] Other areas except sensitive areas are non-sensitive areas. Non-sensitive areas usually have lower risks and are basically areas with large traffic flow. Therefore, the probability of abnormal behavior is not very high, and even if it occurs, it is easy to be discovered and stopped by other personnel in time.

[0038] S2. Acquire image information and / or voice information of persons entering sensitive areas in real time, and analyze whether the persons in the images have abnormal behavior based on the image information and / or voice information;

[0039] Since sensitive areas are prone to abnormal behaviors, real-time monitoring is required to avoid failure to detect abnormal behaviors in a timely manner due to untimely monitoring.

[0040] Exemplary, sensitive areas include the top floor, balcony, bathroom and student dormitory. For the top floor and balcony, real-time monitoring can be performed by installing surveillance cameras; for the top floor area, if there is no need to go to the area, door locks can be added to prohibit school personnel from going to the area. However, bathrooms and student dormitories are not completely public places and have a certain degree of privacy. Therefore, they are often high-incidence areas for bullying. Since it is not convenient to directly install surveillance cameras in these areas, it is easy to form a "hotbed" for bullying classmates. Therefore, in this embodiment, a voice detection device is set in the above areas, and the voice information of the target area is obtained and analyzed to determine whether there is abnormal behavior (mainly bullying behavior) in the target area, such as whether there are insults, screams, etc. The analysis of the voice information can be achieved using a pre-trained voice analysis model, and the samples for model training can be made using existing materials (bullying voice clips). In addition, it is also necessary to prevent the perpetrators from making noise in advance and interfering with the collection of voice information. Therefore, an anti-interference module is also required. After the target person enters the target area (bathroom or student dormitory), if the noise appears for a time greater than the set value (for example, 10 minutes), it is necessary to conduct a strict inspection of the person leaving the target area. For example, the body shape information and facial feature information of the target person can be obtained to determine whether the target person is injured or crying.

[0041] S3. If there is, an alarm operation is performed, and the identity information of the person who committed the abnormal behavior is obtained based on the image information and / or voice information, and the person is marked as a key focus object;

[0042] If abnormal behavior is detected in a sensitive area, an alarm operation must be executed immediately. When teachers, leaders, parents or case handlers receive the alarm information, they need to respond immediately and go to the scene of the incident. At the same time, they can use voice persuasion or deterrence to intervene in the continuation of abnormal behavior.

[0043] S4. Obtain characteristic information of the students entering the school, mainly including facial characteristic information;

[0044] S5. Determine whether the target person is an on-campus or off-campus person based on feature information; teachers, students, leaders and other staff on campus need to enter facial feature information, identity information, etc. in advance.

[0045] S6. If the person is an off-campus staff, the first management method shall be adopted;

[0046] Outsiders are usually booked visitors, such as parents, who need to make an appointment in advance or obtain permission from school teachers or leaders before visiting.

[0047] Specifically, the first management method includes:

[0048] A1. Obtain and record the facial features, time of entry and exit, behavior and movement trajectory of people outside the school;

[0049] A2. Determine whether the person outside the school is suspicious based on the behavior information and the movement trajectory within the school (for example, whether the purpose of his visit is consistent with his movement trajectory; if not, it may be that the person is unfamiliar with the terrain and is lost); if so, call the police and intervene;

[0050] Specifically: if the other party is lost, a receptionist, robot or drone will be dispatched to receive them and guide them to the destination along the correct route; if they have bad behavior, such as arguing, injuring, stealing, etc., specific personnel (such as security guards) or robots or drones will be dispatched to stop or drive them away, and their facial features, behavior information, etc. will be provided to the police station to facilitate the police to judge the case.

[0051] A3. Obtain unexpected events that occurred on campus within a preset time period after the non-school personnel left the school (for example, whether there were any items lost or people lost contact, or were injured, etc.), and analyze the correlation between the non-school personnel and the unexpected events based on the time information, behavior information, and movement trajectory of the non-school personnel entering and leaving the campus. Based on the time information, behavior information, and movement trajectory of the non-school personnel entering and leaving the campus, as well as the time and place of the unexpected events, analyze whether the target personnel have the time and conditions to commit the crime, so as to facilitate the police to solve the case.

[0052] S7. If the target person is an on-campus person, determine whether the target person is a key focus person;

[0053] S8. If yes, adopt the second management method;

[0054] S9. Otherwise, adopt the third management method.

[0055] Specifically, if the identity of the person on campus is a student, the second management method or the third management method includes: obtaining the time information of the student entering and leaving the campus, and synchronizing the time information to the mobile terminal of the student's guardian in real time, so that the student's guardian can understand the student's whereabouts in time, which helps to protect the student's personal safety.

[0056] Exemplarily, in this embodiment, it is assumed that the managed persons in the second management method and the third management method are all students. The second management method includes: calculating the correlation between any two key focus objects, and if the correlation is greater than a preset correlation threshold, the two key focus objects are grouped or managed in different classes.

[0057] The focus of this embodiment is mainly on the perpetrators of violence against others and those with a tendency to self-mutilation (killing). The former is usually a combination of multiple students who commit violence against a single student, and the latter usually has more serious psychological problems. They may be victims of bullying or mental illnesses (such as depression) caused by other reasons (such as family, interpersonal relationships, and academic performance). Therefore, for the former, in order to avoid multiple perpetrators from colluding with each other, they can be divided into classes or groups to reduce the contact between them and reduce the harm to other vulnerable groups; and for the latter, two people with more serious negative emotions contact each other, which is usually easy to affect each other, causing further deterioration of emotions, so contact should be reduced as much as possible; and perpetrators and those with a tendency to self-mutilation (killing) should be kept away from each other. There is a certain probability that the latter is caused by the long-term bullying of the former. Therefore, in order to avoid the latter from suffering secondary harm, they can be divided into classes or groups to reduce the contact between them.

[0058] Specifically, the steps for calculating the correlation G between two focused objects a and b are as follows:

[0059] Acquire images and videos of the campus except classrooms where the focused object a or b is located;

[0060] Randomly extract N frames of images from the image video, and count the number of frames N in which both the focus objects a and b exist. ab , count the number of image frames N0 belonging to class collective activities in N frames of images;

[0061] Correlation G = (N ab -N0) / N(1); the above preset correlation threshold can be set to 30% to 50%.

[0062] Furthermore, the second management method also includes: in a non-sensitive area of ​​the campus, obtaining behavior information of the target person at a first frequency, analyzing whether the target person has abnormal behavior based on the behavior information, and if so, alarming and intervening; the third management method includes: in a non-sensitive area of ​​the campus, obtaining behavior information of the target person at a second frequency, analyzing whether the target person has abnormal behavior based on the behavior information, and if so, alarming and intervening; recording the number of abnormal behaviors of the target person, and if the number of abnormal behaviors exceeds a threshold, marking the target person as a key focus object; wherein the first frequency is higher than the second frequency.

[0063] Since non-sensitive areas are usually less risky and basically belong to areas with a large flow of people, the probability of abnormal behavior is not very high, and even if it occurs, it is easy to be discovered and stopped by other personnel in time. And most of the people on campus will not have abnormal behavior, but there are a large number of people on campus. If all of them are analyzed one by one, it will greatly increase the computing burden of the server and affect the detection efficiency of abnormal behavior. The occurrence of abnormal behavior is usually more likely to occur in the focus objects. Therefore, in this embodiment, the behavior analysis of the focus objects is higher in the non-sensitive areas of the campus than in other non-focus objects on the campus. Exemplarily, the ratio of the first frequency to the second frequency can be set to 2 (2:1) to 4 (4:1). If the second frequency is once every two hours, the first frequency is once an hour to once half an hour.

[0064] In addition, even in non-sensitive areas, abnormal behaviors may occur. According to the type and severity of abnormal behaviors, thresholds can be set. When the number of abnormal behaviors exceeds the threshold, the target will be listed as a key focus. For example, if there is a behavior such as beating students or teachers, the threshold can be set to 0; when the target person is detected to have stereotyped behavior (repeated, fixed, meaningless behavior, such as shaking the body, hitting the head, etc.), the threshold can be set to 1. When it occurs, parents can be notified to send the person to the hospital for psychological counseling or treatment in time.

[0065] Example 2

[0066] This embodiment provides a campus security management system. Figure 2 As shown, the system includes: a division module 1, which is used to divide the campus into sensitive areas and non-sensitive areas; a real-time monitoring module 2, which is arranged in the sensitive area, and is used to obtain image information and / or voice information of people entering the sensitive area in real time, and analyze whether there is abnormal behavior of the people in the image based on the image information and / or voice information; the abnormal behavior includes violent behavior and self-harm behavior; an alarm module 3, which is used to perform an alarm operation when abnormal behavior occurs; a marking module 4, which is used to obtain the identity information of the person with abnormal behavior based on the image information and / or voice information, and mark it as a key focus object; a feature acquisition module 5, which is used to obtain the feature information of the person entering the school; a feature discrimination module 6, which is used to determine whether the target person belongs to the school personnel or the non-school personnel based on the feature information; a management control module 7, which is used to manage the non-school personnel using the first management method, manage the key focus objects using the second management method, and manage the non-key focus objects using the third management method; the second management method includes: calculating the correlation between any two key focus objects, if the correlation is greater than the preset correlation threshold, the two key focus objects are grouped or managed in classes.

[0067] Furthermore, the system also includes: a fixed monitoring component 8, which is arranged in a non-sensitive area, and is used to obtain image information of people in the non-sensitive area, and extract facial feature information and behavior information of the target person based on the image information; a mobile monitoring component 9, which is used to patrol the monitoring blind spots in sensitive areas and the monitoring blind spots in non-sensitive areas, and receive commands from the management control module to go to a specific location to perform monitoring tasks.

[0068] Furthermore, the mobile monitoring component 9 includes: a robot (or a robot dog) or a drone, on which an intervention module is installed, and the intervention module is used to intervene in the abnormal behavior of the target person (the intervention method includes voice admonition or warning, water flow suppression, strong flash, etc., which is selected according to the adaptability of the occasion); if the target person has abnormal behavior, the management control module is also used to obtain the location of the target person, and drive the robot or drone to the location of the target person to intervene in the abnormal behavior of the target person.

[0069] Furthermore, in an embodiment of the present application, the robot or drone includes two working modes: a daily inspection mode and a target tracking mode; the daily inspection mode is: the robot or drone regularly performs monitoring tasks according to a pre-planned fixed inspection route; the target tracking mode is: when abnormal behavior is found, the abnormal person is tracked and the abnormal person's route and image information are recorded until a command to stop tracking is received or the abnormal person leaves the monitoring range of the robot or drone.

[0070] Exemplary methods of tracking abnormal behavior by drone include:

[0071] B1. Obtain the coordinates (x0, y0, z0) of the person who committed the abnormal behavior in the pre-established three-dimensional coordinate system

[0072] B2. Establish the three-dimensional kinematic model of the UAV: In formula (2), x, y, z are the position coordinates of the UAV, v is the speed of the UAV, α is the heading angle of the UAV, β is the track angle of the UAV, and ω is the control input (desired heading angle);

[0073] B3. Set the tracking and guidance law of the UAV: In the above formula (3), θ represents the angle between the speed of the drone and the relative distance, and the relative distance is the relative distance between the drone and the abnormal person, where: λ represents the straight-line distance between the drone and the abnormal actor, and its expression is: L represents the expected distance between the drone and the abnormal person, which is set to be known. In this embodiment, fixed-distance tracking is used to ensure the stability of shooting; h represents the height of the drone, where h=z=vsinβ(6); H represents the expected height value between drones, and H=h0+ΔH(7), where h0 represents the height of the abnormal person, and ΔH represents the fixed height difference between the abnormal person and the drone; μ1 and μ2 are adjustment parameters, which can be customized;

[0074] B4. Calculate the control variable (ω) according to the three-dimensional kinematic model, the tracking and guidance law of the UAV, and equations (4), (5), (6), and (7); the control variable includes the desired heading angle of the UAV, and the control variable is updated in real time as the position of the abnormal actor moves;

[0075] B5. Update the control quantity and control the drone to track the person with abnormal behavior.

[0076] Furthermore, in an embodiment of the present application, the sensitive area includes a bathroom, and the real-time monitoring module 2 provided in the bathroom includes a voice detection unit, which is used to obtain voice information in the bathroom and analyze whether there is abnormal behavior in the bathroom based on the voice information.

[0077] The system provided in the embodiment of the present application can be used to execute the method in the above-mentioned embodiment 1. For details, please refer to embodiment 1 and will not be repeated here.

[0078] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only the specific implementation method of the present application and is not intended to limit the scope of protection of the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A campus safety management method, characterized in that: The method comprises: Identify sensitive areas on campus; Acquire image information and / or voice information of persons entering sensitive areas in real time, and analyze whether the persons in the images have abnormal behaviors based on the image information and / or voice information; the abnormal behaviors include violent behaviors and self-harming behaviors; If there is, an alarm operation is performed, and the identity information of the abnormal person is obtained based on the image information and / or voice information, and the person is marked as a key focus object; Obtain characteristic information of students entering the school; Determine whether the target person is an on-campus person or an off-campus person based on feature information; If the person is an off-campus staff, the first management method shall be adopted; If the target person is an on-campus person, determine whether he or she is a key focus person; if so, adopt the second management method; otherwise, adopt the third management method; The second management method includes: calculating the correlation between any two key attention objects, and if the correlation is greater than a preset correlation threshold, grouping or class management is performed on the two key attention objects.

2. The campus safety management method according to claim 1, characterized in that: The identified sensitive areas on campus include: Collect statistics on the areas where abnormal behaviors occurred in the history of campus; Areas where historical abnormal behaviors have occurred more than A times are identified as sensitive areas of the campus, and the rest of the areas are non-sensitive areas; A is the set value.

3. The campus safety management method according to claim 1, characterized in that: The first management method comprises: Obtain and record the facial features, time of entry and exit, behavior and movement trajectory of non-school personnel; Determine whether the person outside the school is suspicious based on the behavior information and the movement trajectory within the school; if so, call the police and intervene; Obtain unexpected events that occur on campus within a preset time period after the non-school personnel leave the school, and analyze the correlation between the non-school personnel and the unexpected events based on the time information, behavior information and movement trajectory of the non-school personnel entering and leaving the campus.

4. The campus safety management method according to claim 1, characterized in that: In the second management method, the steps of calculating the correlation G between two focused objects a and b are as follows: Acquire images and videos of the campus except classrooms where the focused object a or b is located; Randomly extract N frames of images from the image video, and count the number of frames N in which both the focus objects a and b exist. ab , count the number of image frames N0 belonging to class collective activities in N frames of images; Correlation G = (N ab -N0) / N(1).

5. The campus safety management method according to claim 1, characterized in that: The second management method includes: obtaining behavior information of a target person at a first frequency in a non-sensitive area of ​​the campus, analyzing whether the target person has abnormal behavior based on the behavior information, and if so, alarming and intervening; The third management method includes: obtaining the behavior information of the target person at the second frequency in the non-sensitive area of ​​the campus, analyzing whether the target person has abnormal behavior according to the behavior information, and if so, alarming and intervening; recording the number of abnormal behaviors of the target person, and if the number of abnormal behaviors exceeds a threshold, marking the target person as a key focus object; The first frequency is higher than the second frequency.

6. A campus safety management system, characterized in that: The system comprises: Division module, used to divide the campus into sensitive and non-sensitive areas; A real-time monitoring module is provided in the sensitive area, and is used to obtain image information and / or voice information of persons entering the sensitive area in real time, and analyze whether the persons in the image have abnormal behavior based on the image information and / or voice information; the abnormal behavior includes violent behavior and self-harming behavior; An alarm module is used to execute an alarm operation when abnormal behavior occurs; A marking module, used to obtain the identity information of the person who has abnormal behavior based on the image information and / or voice information, and mark him as a key focus object; A feature acquisition module is used to obtain feature information of people entering the school; A feature identification module is used to determine whether the target person is an on-campus person or an off-campus person based on feature information; A management control module is used to manage off-campus personnel using a first management method, manage key focus objects using a second management method, and manage non-key focus objects using a third management method; the second management method includes: calculating the correlation between any two key focus objects, and if the correlation is greater than a preset correlation threshold, the two key focus objects are grouped or managed in different classes.

7. The campus safety management system according to claim 6, characterized in that: The system further comprises: A fixed monitoring component is set in a non-sensitive area to obtain image information of people in the non-sensitive area and extract facial feature information and behavior information of the target person based on the image information; The mobile monitoring component is used to inspect the monitoring blind spots in sensitive areas and the monitoring blind spots in non-sensitive areas, and receive commands from the management and control module to go to specific locations to perform monitoring tasks.

8. The campus safety management system according to claim 7, characterized in that: The mobile monitoring component includes: a robot or a drone, on which an intervention module is installed, and the intervention module is used to intervene in the abnormal behavior of the target person; If the target person has abnormal behavior, the management control module is also used to obtain the location of the target person and drive the robot or drone to the location of the target person to intervene in the abnormal behavior of the target person.

9. The campus safety management system according to claim 8, characterized in that: The robot or drone includes two working modes: daily inspection mode and target tracking mode; The daily inspection mode is: to make the robot or drone regularly perform monitoring tasks according to the pre-planned fixed inspection route; The target tracking mode is: when abnormal behavior is found, the abnormal person is tracked and the abnormal person's route and image information are recorded until a command to stop tracking is received or the abnormal person leaves the monitoring range of the robot or drone.

10. The campus safety management system according to claim 9, characterized in that: The target tracking mode execution steps of the UAV include: B1. Obtain the coordinates (x0, y0, z0) of the person with abnormal behavior in a pre-established three-dimensional coordinate system; B2. Establish the three-dimensional kinematic model of the UAV: In formula (2), x, y, z are the position coordinates of the UAV, v is the speed of the UAV, α is the heading angle of the UAV, β is the track angle of the UAV, and ω is the control variable; B3. Set the tracking and guidance law of the UAV: In the above formula (3), θ represents the angle between the speed of the drone and the relative distance, and the relative distance is the relative distance between the drone and the abnormal person, where: λ represents the straight-line distance between the drone and the abnormal actor, and its expression is: L represents the expected distance between the UAV and the abnormal actor, which is set to be known; h represents the height of the UAV, where h = z = vsinβ (6); H represents the expected height value of the UAV, and H = h0 + ΔH (7), where h0 represents the height of the abnormal actor, and ΔH represents the fixed height difference between the abnormal actor and the UAV; μ1 and μ2 are adjustment parameters; B4. Calculate the control quantity according to the three-dimensional kinematic model, the tracking guidance law of the UAV and equations (4), (5), (6) and (7); the control quantity includes the desired heading angle of the UAV; B5. Update the control quantity and control the drone to track the person with abnormal behavior.

Images