Intelligent security AI early warning system

Abstract

The invention discloses an intelligent security AI early warning system, and belongs to the technical field of intelligent security, the early warning system analyzes the motion path of external personnel in a monitoring area, and when the external personnel frequently appear on the path with less personnel activity amount in the motion process in the monitoring area, the related workers are prompted in time, so that the safety of the external personnel is improved. And the monitoring level of the external personnel is improved, so that early warning is realized, and occurrence or expansion of man-made accidents is avoided. In addition, in consideration of two typical conditions of people who invade the monitoring area abnormally, on the basis of the two typical conditions, the movement speeds of high-risk people in different single advancing paths are further analyzed; therefore, whether the overall movement condition of the high-risk personnel accords with the overall portrait condition of the personnel in the white list in the monitoring area is judged, so that whether the early warning level of the corresponding high-risk personnel needs to be improved is further judged, the false alarm probability can be further reduced, and warning resources are reasonably utilized.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent security technology, and specifically relates to an intelligent security AI early warning system. Background Art

[0002] Smart security early warning is a system that uses high-tech means to achieve real-time monitoring, analysis and instant alarm of potential security threats. Smart security early warning plays an important role in maintaining social stability and protecting people's lives and property. Compared with traditional security, smart security has the characteristics of digitalization and integration. It uses the latest information technology, including artificial intelligence, big data, cloud computing, etc., to achieve a leap from simple monitoring to comprehensive early warning.

[0003] Intelligent security warning plays an irreplaceable role in improving the level of social security management and preventing and reducing safety accidents. With the continuous advancement of technology and the continuous expansion of its application scope, intelligent security warning will become more intelligent and precise, and better serve social development and people's needs.

[0004] In the prior art, common intelligent security warnings are to recognize the faces of people entering the monitoring area. When a non-white-listed person enters the monitoring area or a red-listed person enters the monitoring area, an alarm is initiated to prompt the monitoring personnel to perform subsequent operations such as expulsion and monitoring to avoid accidents such as theft and injury. However, this method is less effective in areas with complex personnel composition, is prone to frequent false alarms, and cannot achieve accurate security warnings. In order to solve the above problems, the present invention provides the following technical solutions. Summary of the invention

[0005] The purpose of the present invention is to provide an intelligent security AI early warning system to solve the problem that the existing technology of judging whether people entering the monitoring area need to be warned by means of white lists, red lists, etc. is prone to false alarms and cannot achieve accurate security early warnings.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] An intelligent security AI early warning system, comprising:

[0008] A monitoring module, including monitoring cameras distributed in the monitoring area;

[0009] Access control module, used to collect and compare information about people entering and leaving the monitoring area;

[0010] A memory device for storing the identity information of whitelisted personnel in the monitored area and temporarily storing the identity information of external personnel;

[0011] The backend processing center is used to determine whether an early warning is needed;

[0012] The method for the background processing center to determine whether an early warning is needed includes:

[0013] Step 1: Obtain path nodes in the monitoring area; mark a path between two adjacent path nodes as a single travel path;

[0014] Step 2: obtaining a prosperity coefficient F of each single travel path; the prosperity coefficient F can represent the average number of people moving on each single travel path;

[0015] For a single moving path, when the corresponding prosperity coefficient F is greater than or equal to F1, the corresponding single moving path is considered to be a regular moving path; on the contrary, when the corresponding prosperity coefficient F is less than F1, the corresponding single moving path is considered to be an unconventional moving path; wherein F1 is a preset coefficient;

[0016] Step 3: When the access control module recognizes that an outsider has entered the monitoring area, the movement path of the outsider in the monitoring area is obtained;

[0017] Step 4: When α=q1 / q2≥α1 and q2≥q3 is established, the corresponding external personnel are considered to be high-risk personnel, and the background processing center sends a command to the alarm unit;

[0018] Wherein, q1 is the sum of the number of times the motion path of the external personnel covers various unconventional motion paths, q2 is the sum of the number of times the motion path of the external personnel covers all single travel paths; q3 is a preset number value, and α1 is a preset proportional coefficient.

[0019] Furthermore, the calculation method of the prosperity coefficient F is:

[0020] For a single travel path, within a monitoring cycle, a frame image covering the single travel path is acquired every preset time t1 to obtain a plurality of frame images;

[0021] Get the number of people r walking on a single path in each frame image;

[0022] Get the average value rp of the number of people r in these several frame images;

[0023] Obtain typical values ​​of several corresponding average values ​​rp in several past monitoring cycles, and use the typical value as the prosperity coefficient F corresponding to a single travel path.

[0024] Furthermore, the alarm information includes the current location of the high-risk person, identity feature information of the high-risk person, and other information that can be used to identify and locate the high-risk person.

[0025] Further, a monitoring cycle is divided into a number of monitoring periods;

[0026] Obtain the prosperity coefficient of each single travel path in each monitoring period;

[0027] When an outsider enters the monitoring area, the monitoring period in which he is located and the prosperity coefficient F corresponding to the monitoring period are obtained;

[0028] For a single moving path, when its corresponding prosperity coefficient F is greater than or equal to F1, the corresponding single moving path is considered to be a regular moving path during the corresponding monitoring period; conversely, when its corresponding prosperity coefficient F is less than F1, the corresponding single moving path is considered to be an unconventional moving path during the corresponding monitoring period;

[0029] When α=q1 / q2≥α1 and q2≥q3 are established, the corresponding external personnel are considered to be high-risk personnel, and the background processing center sends an instruction to the alarm unit.

[0030] Furthermore, when the time that the external person is in the monitoring area spans two or more monitoring periods, the prosperity coefficient compared with F1 is the average of the prosperity coefficients of the corresponding two or more monitoring areas.

[0031] Further, after the high-risk personnel are determined, the single travel duration t2i of the high-risk personnel in each unconventional movement path covered by the movement route is obtained, where 1≤i≤m, and m is the sum of the number of times each unconventional movement path is covered by the movement route of the high-risk personnel;

[0032] Get the average time t3i corresponding to each unconventional motion path;

[0033] According to βi=t2i / t3i, m corresponding proportional coefficients βi are calculated;

[0034] Obtain the typical value β of the m proportional coefficients βi;

[0035] When β≥βy1 or β≤βy2 is satisfied, the corresponding high-risk personnel are considered abnormal and the warning level is further increased; βy1 is a preset value greater than 1, and βy2 is a preset value less than 1.

[0036] Furthermore, the calculation method of t3i is:

[0037] For an unconventional motion path, obtain a single travel time t4 of other people moving on the unconventional motion path;

[0038] After obtaining several single travel durations t4, the average value of these several single travel durations t4 is obtained, and the average value is t3i

[0039] Beneficial effects of the present invention:

[0040] 1. The present invention analyzes the movement paths of outsiders in the monitoring area. When outsiders frequently appear on paths with less human activity during their movement in the monitoring area, it is assumed that the outsiders intentionally avoid the crowds and there may be abnormalities. The relevant staff will be reminded in time to increase the monitoring level of the outsiders, thereby achieving early warning and avoiding the occurrence or expansion of man-made accidents.

[0041] 2. The present invention discusses different time periods separately, taking into account the fact that the flow of people in many areas has a large periodic regularity, thereby significantly improving the accuracy of early warning.

[0042] 3. The present invention takes into account that there are two typical situations of people who abnormally invade the monitoring area. On this basis, this embodiment further analyzes the movement speed of high-risk personnel in different single travel paths, so as to judge whether the overall movement of high-risk personnel is too slow or too fast, and whether it does not conform to the overall portrait of the white-list personnel in the monitoring area. In this way, it is further judged whether it is necessary to increase the warning level of the corresponding high-risk personnel, which can further reduce the probability of false alarms, reasonably utilize warning resources, and improve the accuracy of warnings. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The present invention will be further described below in conjunction with the accompanying drawings.

[0044] Figure 1 It is a schematic diagram of the framework structure of an intelligent security AI early warning system;

[0045] Figure 2 It is a flowchart of the method by which the background processing center determines whether an early warning is needed. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not 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.

[0047] Example 1

[0048] An intelligent security AI early warning system, such as Figure 1 As shown, including:

[0049] The monitoring module includes monitoring cameras distributed in the monitoring area. The monitoring range of the monitoring cameras can cover the monitoring area. The image information collected by the monitoring cameras is transmitted to the background processing center.

[0050] Access control module, used to collect and compare information about people entering and leaving the monitoring area;

[0051] A memory device for storing identity information of whitelisted personnel in the monitoring area, and for temporarily storing identity information of external personnel who temporarily enter and exit the monitoring area;

[0052] The background processing center is used to analyze the movement trajectory of external personnel in the monitoring area, and determine whether an early warning is needed based on the movement trajectory, and remind relevant staff to be vigilant against external personnel.

[0053] like Figure 2 As shown, the method for the background processing center to determine whether an early warning is needed includes the following steps:

[0054] Step 1: Create a plan view (such as a top view) of the monitoring area;

[0055] Acquire path nodes in the monitoring area, wherein the path nodes include intersections of paved roads in the monitoring area, intersections between paved roads and roads formed by human trampling, and intersections between roads formed by human trampling;

[0056] Mark a path between two adjacent path nodes as a single travel path;

[0057] There can be two or more single travel paths between two adjacent path nodes;

[0058] Step 2, obtaining the prosperity coefficient F of each single travel path;

[0059] The prosperity coefficient F can represent the average number of people moving on each single path;

[0060] For a single moving path, when its corresponding prosperity coefficient F is greater than or equal to F1, the corresponding single moving path is considered to be a regular moving path;

[0061] On the contrary, when the corresponding prosperity coefficient F is less than F1, the corresponding single moving path is considered to be an unconventional moving path;

[0062] Where F1 is a preset coefficient, and F is the critical value, which can represent the number of people moving on a single path;

[0063] In one embodiment of the present invention, the calculation method of the prosperity coefficient F is:

[0064] For a single travel path, within a monitoring period, a frame image covering the single travel path is obtained every preset time t1, thereby obtaining a plurality of frame images; generally speaking, a monitoring period refers to a day or a part of a day (for example, in a certain community, the time between 6 am and 10 pm is taken as a monitoring period);

[0065] Get the number of people r walking on a single path in each frame image;

[0066] Get the average value rp of the number of people r in these several frame images;

[0067] Obtain typical values ​​of several corresponding average values ​​rp in several past monitoring cycles, and use the typical values ​​as the prosperity coefficient F corresponding to a single travel path;

[0068] Among them, the typical value is the average value, median or average value of the remaining rp after deleting some values ​​with large deviations, etc., which can represent the characteristics of the group of several average rp values;

[0069] Step 3: When the access control module recognizes that a person who does not belong to the whitelist corresponding to the monitoring area enters the monitoring area, the access control module transmits the identity feature information of the person to the storage device, and for the convenience of description, marks the person as an outsider;

[0070] The identity feature information includes at least one of the information such as facial image information and clothing of a person that can identify the identity of the person;

[0071] The monitoring module obtains the action image of the external person in the monitoring area, and obtains the movement path of the external person in the monitoring area according to the action image of the external person in the monitoring area;

[0072] Step 4: According to the movement path of the external personnel, the single travel path covered by the external personnel when moving in the monitoring area and the number of times each single travel path is covered are obtained;

[0073] When α=q1 / q2≥α1 and q2≥q3 are true, the corresponding external personnel are considered to be high-risk personnel, and the background processing center sends instructions to the alarm unit, which sends the alarm information to the corresponding staff;

[0074] Where q1 is the sum of the number of times the movement path of the external personnel covers all unconventional movement paths, q2 is the sum of the number of times the movement path of the external personnel covers all single movement paths; q3 is the preset number value, and α1 is the preset proportional coefficient;

[0075] The alarm information includes the current location of the high-risk person, the identity feature information of the high-risk person, and other information that can be used to identify and locate the high-risk person.

[0076] The present invention analyzes the movement paths of outsiders in the monitoring area. When outsiders frequently appear on paths with less human activity during their movement in the monitoring area, it is assumed that the outsiders intentionally avoid the crowd and there may be an abnormality. The relevant staff will be reminded in time to increase the monitoring level of the outsiders, thereby achieving early warning and avoiding the occurrence or expansion of man-made accidents.

[0077] Example 2

[0078] Based on Example 1, a monitoring cycle is divided into several monitoring time periods;

[0079] Obtain the prosperity coefficient F of each single travel path in each monitoring period;

[0080] The specific calculation method is:

[0081] For a single travel path, in the past n monitoring cycles, taking the same monitoring period as an example, a frame image covering the single travel path is acquired at a preset time interval in the monitoring period, thereby acquiring a plurality of frame images;

[0082] Get the number of people r1 walking on a single path in each frame image;

[0083] Get the average value rpp of the number of people r1 in these several frame images;

[0084] Obtain typical values ​​of several average values ​​rpp corresponding to the monitoring period in several past monitoring cycles, and use the typical value as the prosperity coefficient F of the corresponding single travel path in the corresponding monitoring period;

[0085] When an outsider enters the monitoring area, the monitoring period in which he is located and the prosperity coefficient F corresponding to the monitoring period are obtained;

[0086] For a single moving path, when its corresponding prosperity coefficient F is greater than or equal to F1, the corresponding single moving path is considered to be a regular moving path during the corresponding monitoring period;

[0087] On the contrary, when the corresponding prosperity coefficient F is less than F1, the corresponding single moving path is considered to be an unconventional moving path during the corresponding monitoring period;

[0088] It should be noted that when the time an outsider spends in the monitoring area spans two or more monitoring periods, the prosperity coefficient compared with F1 is the average of the prosperity coefficients of the corresponding two or more monitoring areas;

[0089] Similarly, according to the movement path of the external personnel, the single travel path covered by the external personnel when moving in the monitoring area and the number of times each single travel path is covered are obtained;

[0090] When α=q1 / q2≥α1 and q2≥q3 are established, the corresponding external personnel are considered to be high-risk personnel, and the background processing center sends an instruction to the alarm unit, and the alarm unit sends the alarm information to the corresponding staff.

[0091] Since the flow of people in many areas has a large periodic regularity, the accuracy of early warning can be significantly improved by discussing different time periods separately.

[0092] Example 3

[0093] On the basis of Embodiment 1 and Embodiment 2, after the high-risk personnel are determined, the single travel duration t2i of the high-risk personnel in each unconventional motion path covered by the motion route can be further obtained (that is, the time taken by the high-risk personnel to move from one path node of the unconventional motion path to another path node, and since the high-risk personnel may walk the same unconventional motion path twice or more, one unconventional motion path may correspond to two or more travel durations t2), where 1≤i≤m, m is the sum of the number of unconventional motion paths covered by the motion route of the high-risk personnel;

[0094] Get the average time t3i corresponding to each unconventional motion path;

[0095] According to βi=t2i / t3i, m corresponding proportional coefficients βi are calculated;

[0096] Obtain the typical value β of the m proportional coefficients βi;

[0097] When β≥βy1 or β≤βy2 is satisfied, it is considered that the corresponding high-risk personnel are abnormal, and the warning level is further improved on the basis of Example 1 and Example 2.

[0098] Wherein, βy1 is a preset value greater than 1, and βy2 is a preset value less than 1.

[0099] The calculation method of t3i is:

[0100] For an unconventional motion path, obtain a single travel time t4 of other people moving on the unconventional motion path;

[0101] After obtaining a number of single travel durations t4, an average value of the number of single travel durations t4 is obtained, and the average value is t3i.

[0102] The present invention takes into account that there are two typical situations of people who abnormally invade the monitoring area. One is that they have a clear goal and will rush to the destination at a faster speed, and the other is that they will choose to observe the destination. On this basis, this embodiment further analyzes the movement speed of high-risk personnel in different single travel paths, so as to judge whether the overall movement of high-risk personnel is too slow or too fast, and whether it does not conform to the overall portrait of the white-list personnel in the monitoring area, so as to further judge whether it is necessary to improve the warning level of the corresponding high-risk personnel, which can further reduce the probability of false alarms, reasonably utilize warning resources, and improve the accuracy of warnings.

[0103] The above contents are merely examples and explanations of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the invention or exceed the scope defined by the claims, they shall all fall within the protection scope of the present invention.

Claims

1. An intelligent security AI early warning system, characterized in that: include: A monitoring module, including monitoring cameras distributed in the monitoring area; Access control module, used to collect and compare information about people entering and leaving the monitoring area; A memory device for storing the identity information of whitelisted personnel in the monitored area and temporarily storing the identity information of external personnel; The backend processing center is used to determine whether an early warning is needed; The method for the background processing center to determine whether an early warning is needed includes: Step 1: Obtain path nodes in the monitoring area; mark a path between two adjacent path nodes as a single travel path; Step 2: obtaining a prosperity coefficient F of each single travel path; the prosperity coefficient F can represent the average number of people moving on each single travel path; For a single moving path, when the corresponding prosperity coefficient F is greater than or equal to F1, the corresponding single moving path is considered to be a regular moving path; on the contrary, when the corresponding prosperity coefficient F is less than F1, the corresponding single moving path is considered to be an unconventional moving path; wherein F1 is a preset coefficient; Step 3: When the access control module recognizes that an outsider has entered the monitoring area, the movement path of the outsider in the monitoring area is obtained; Step 4: When α=q1 / q2≥α1 and q2≥q3 is established, the corresponding external personnel are considered to be high-risk personnel, and the background processing center sends a command to the alarm unit; Wherein, q1 is the sum of the number of times the motion path of the external personnel covers various unconventional motion paths, q2 is the sum of the number of times the motion path of the external personnel covers all single travel paths; q3 is a preset number value, and α1 is a preset proportional coefficient.

2. According to claim 1, the intelligent security AI early warning system is characterized in that: The calculation method of the prosperity coefficient F is: For a single travel path, within a monitoring cycle, a frame image covering the single travel path is acquired every preset time t1 to obtain a plurality of frame images; Get the number of people r walking on a single path in each frame image; Get the average value rp of the number of people r in these several frame images; Obtain typical values ​​of several corresponding average values ​​rp in several past monitoring cycles, and use the typical value as the prosperity coefficient F corresponding to a single travel path.

3. The intelligent security AI early warning system according to claim 1 is characterized in that: The alarm information includes the current location of the high-risk person, the identity feature information of the high-risk person, and other information that can be used to identify and locate the high-risk person.

4. The intelligent security AI early warning system according to claim 1 is characterized in that: Divide a monitoring cycle into several monitoring periods; Obtain the prosperity coefficient of each single travel path in each monitoring period; When an outsider enters the monitoring area, the monitoring period in which he is located and the prosperity coefficient F corresponding to the monitoring period are obtained; For a single moving path, when its corresponding prosperity coefficient F is greater than or equal to F1, the corresponding single moving path is considered to be a regular moving path during the corresponding monitoring period; On the contrary, when the corresponding prosperity coefficient F is less than F1, the corresponding single moving path is considered to be an unconventional moving path during the corresponding monitoring period; When α=q1 / q2≥α1 and q2≥q3 are established, the corresponding external personnel are considered to be high-risk personnel, and the background processing center sends an instruction to the alarm unit.

5. The intelligent security AI early warning system according to claim 4 is characterized in that: When the time that an outsider spends in the monitoring area spans two or more monitoring periods, the prosperity coefficient compared with F1 is the average of the prosperity coefficients of the corresponding two or more monitoring areas.

6. The intelligent security AI early warning system according to claim 1 is characterized in that: After the high-risk personnel are determined, the single travel duration t2i of the high-risk personnel in each unconventional movement path covered by the movement route is obtained, where 1≤i≤m, and m is the sum of the number of times each unconventional movement path is covered by the movement route of the high-risk personnel; Get the average time t3i corresponding to each unconventional motion path; According to βi=t2i / t3i, m corresponding proportional coefficients βi are calculated; Obtain the typical value β of the m proportional coefficients βi; When β≥βy1 or β≤βy2 is satisfied, the corresponding high-risk personnel are considered abnormal and the warning level is further increased; βy1 is a preset value greater than 1, and βy2 is a preset value less than 1.

7. The intelligent security AI early warning system according to claim 6 is characterized in that: The calculation method of t3i is: For an unconventional motion path, obtain a single travel time t4 of other people moving on the unconventional motion path; After obtaining a number of single travel durations t4, an average value of the number of single travel durations t4 is obtained, and the average value is t3i.

Images