Community visitor management method and system based on AI technology

Abstract

The invention relates to the technical field of security and protection, in particular to a community visitor management method and system based on the AI technology, and the method comprises the steps: employing a front-end interaction device to obtain a first face image of a visitor, employing a matrix pressure sensing device to monitor a pressure matrix of the visitor to the ground in real time, and carrying out the recognition of the first face image of the visitor based on a face recognition algorithm; matching the first face image with an image in a pre-registered face image library, if the matching succeeds, sending an opening instruction to the access control equipment, recording access control opening time, then calculating a pressure gradient difference and a module value of a pressure matrix after the access control opening time, and calculating the pressure gradient difference and the module value of the pressure matrix; lastly, continuously judging whether any one of the following conditions is met: the pressure gradient difference is less than or equal to 0; the pressure gradient difference is greater than 0 and the modulus value is less than the pressure threshold. If not, a following alarm is given out. Compared with the prior art, the method provided by the invention solves the problem of failure of the trailing alarm in the outdoor environment.

Description

Technical Field

[0001] The present invention relates to the field of security technology, and in particular to a community visitor management method and system based on AI technology. Background Art

[0002] The community visitor management system is an intelligent system specially used to manage the entry and exit of community visitors, which can improve the security of the community. It is generally composed of front-end interactive equipment, access control equipment, back-end management equipment and mobile application terminals. Before non-community residents apply to enter the community, they generally need to use the mobile application terminal to register the visit information to the back-end management device, and then use the AI ​​face recognition technology to determine whether it is the person in the front-end interactive device. If so, the gate of the access control device is opened to allow the visitor to pass. At the moment when the visitor passes through the access control device, it is usually detected by multiple groups of infrared sensors distributed in the upper, middle and lower positions and installed on the inside of the access control device to detect whether a single person passes. If a single person passes, the infrared signal detection feature has only one section of infrared distance detection data, and the visitor is safely released at this time. If there are multiple sections of distance detection data in the time sequence, a tailing alarm is issued, indicating that a suspicious person is following the visitor into the community at this time.

[0003] However, infrared sensors have obvious defects when used in outdoor environments and cannot meet the actual needs of outdoor environments. Specifically, the outdoor environment is complex and changeable. Factors such as direct sunlight, wind-blown tree shadows, smoke and dust may interfere with infrared signals, causing the effective detection distance of infrared sensors to change and affecting the accuracy of sensors. When suspicious persons follow visitors into the community, infrared sensors are affected by these external factors and often fail to detect suspicious persons, causing the following alarm to fail, which is a major hidden danger to community safety. Summary of the invention

[0004] The present invention provides a community visitor management method and system based on AI technology, which are used to solve the problem of failure of tailgating alarms in outdoor environments in the prior art.

[0005] In a first aspect, the present invention discloses a community visitor management method based on AI technology, which is used for a community visitor management system with a matrix pressure sensor device installed outdoors. The method of the present invention comprises: When the front-end interactive device is used to obtain the first face image of the visitor, a matrix pressure sensor device is simultaneously used to monitor the pressure matrix of the visitor on the ground in real time; Based on the face recognition algorithm, the first face image is matched with the images in the pre-registered face image library. If the match is successful, an opening instruction is sent to the access control device and the access control opening time is recorded; Calculate the pressure gradient difference and modulus of the pressure matrix after the access control opening time; the calculation formula for the pressure gradient difference is:

[0006] in, , The access control opening time. is the pressure gradient difference, For time Pressure Matrix The pressure gradient, For time Pressure Matrix The pressure gradient; Continuously determine whether any of the following conditions are met: the pressure gradient difference is less than or equal to 0; the pressure gradient difference is greater than 0 and the modulus value is less than the pressure threshold; If none of the conditions are met, a tailing alarm will be issued.

[0007] Beneficial effects: The method of the present invention is specifically used in outdoor environments. The method first uses a parallel process to obtain the first face image and pressure matrix in real time. At the moment after the face recognition algorithm recognizes and issues the access control, the pressure gradient difference and modulus value are continuously calculated. If the pressure gradient difference is less than or equal to 0 after the access control is opened, it means that a single person can pass normally or a device with a linearly increasing pressure can pass. If the pressure gradient difference is greater than 0 and the modulus value is less than the pressure threshold, it means that a single person is passing with a child. If both of the above conditions are not met, it means that there is a possibility of multiple people following, and a tailing alarm is issued. Compared with the prior art, the method of the present invention uses a matrix pressure sensing device to obtain tailing-related detection information, and uses the analysis of the pressure matrix data to determine whether there is a tailing behavior after the access control is opened. This method is less affected by the interference factors of the outdoor environment, and solves the problem that the existing infrared sensing technology is prone to tailing detection failure in outdoor environments.

[0008] Preferably, time Pressure Matrix The pressure gradient is calculated as:

[0009] In the formula, Represents the pressure matrix exist The pressure gradient at that moment, Indicated in The pressure matrix of the moment, Indicated in The pressure matrix of the moment, express arrive time interval.

[0010] Preferably, the calculation formula of the modulus value is:

[0011] In the formula, For The modulus of the moment pressure matrix, is the number of rows in the pressure matrix, is the number of columns in the pressure matrix.

[0012] Preferably, before calculating the pressure gradient difference and the modulus value of the pressure matrix, the method of the present invention further comprises: like ,Will is replaced with 1.

[0013] Beneficial effect: normalizing the pressure values ​​greater than 0 in the pressure matrix to eliminate interference caused by different weights of different visitors is beneficial to data analysis of the pressure matrix, thereby improving the accuracy and reliability of the present invention for tailgating judgment.

[0014] Preferably, the calculation formula of the pressure threshold is:

[0015] In the formula, is the pressure threshold, is the number of rows in the pressure matrix, is the number of columns of the pressure matrix, The time when the door is opened The pressure matrix, is the proportionality coefficient.

[0016] Beneficial effect: Through the above algorithm design, the pressure threshold will be adaptively adjusted according to the pressure matrix characteristics of the visitor who has undergone face recognition to improve the accuracy of tailgating detection.

[0017] Preferably, the proportionality factor The value range is 0.25-0.5.

[0018] Preferably, the face recognition algorithm uses a multi-frame recognition algorithm to match the first face image with images in a pre-registered face image library, specifically: Calculate the single frame matching degree between each frame of the first face image and the second face image in the face image library; Calculate the matching mean of multiple single-frame matching degrees; Determine whether the matching mean is higher than the matching threshold. If so, the match is successful.

[0019] Beneficial effect: By continuously shooting the facial information of visitors, obtaining the first facial images of multiple consecutive frames, and calculating the matching degree between the multiple first facial images and the second facial images, the accuracy of facial recognition can be improved and the problem of insufficient recognition accuracy of single-frame capture can be avoided.

[0020] Preferably, before using the front-end interactive device to obtain the first facial image of the visitor, the method of the present invention further includes: Identify whether the visitor is in the visitor channel. If so, obtain the visitor's first facial image; if not, send a service suspension pop-up window to the visitor.

[0021] Beneficial effects: The method of the present invention can weakly guide visitors to the visitor channel for face recognition and pressure monitoring. This method realizes channel dedicating and reduces the impact of visitors' visit to the community on the daily entry and exit of local community residents.

[0022] Preferably, the first face image is matched with images in a pre-registered face image library. If the match fails, the method of the present invention further comprises: Read the temporary access request from the mobile application terminal; Verify whether the information in the temporary access request contains the identity ID, access area and contact number. If so, send an opening command to the access control device.

[0023] Beneficial effect: If the visitor has not registered his / her facial information in advance, the access control device can be opened by applying for a temporary access request, which improves the flexibility and convenience of the method of the present invention.

[0024] In a second aspect, the present invention also provides a community visitor management system based on AI technology, comprising a processor and a memory, wherein the memory stores computer program instructions for implementing a community visitor management method based on AI technology, and when the computer program instructions are executed by the processor, the community visitor management method based on AI technology of the first aspect is implemented.

[0025] The beneficial effects of the present invention are: (1) Compared with the prior art, the method of the present invention adopts a matrix pressure sensing device to obtain tailgating-related detection information, and uses the analysis of the pressure matrix data to determine whether there is tailgating behavior after the door is opened. This method is less affected by interference factors in the outdoor environment, and solves the problem that the existing infrared sensing technology is prone to tailgating detection failure in the outdoor environment.

[0026] (2) Compared with the prior art, the method of the present invention normalizes the pressure values ​​greater than 0 in the pressure matrix to eliminate the interference caused by the different weights of different visitors, which is beneficial to data analysis of the pressure matrix and further improves the accuracy and reliability of the present invention for tailgating judgment.

[0027] (3) Compared with the prior art, the pressure threshold of the method of the present invention is adaptively adjusted according to the visitor's own pressure matrix characteristics to improve the accuracy of tailgating detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a flow chart of a community visitor management method based on AI technology in Embodiment 1 of the present invention; Figure 2 This is a diagram showing changes in the pressure matrix of a single person passing through the access control device in the first embodiment of the present invention; Figure 3 This is a diagram showing changes in the pressure matrix of a suitcase passing through a door access control device in the first embodiment of the present invention; Figure 4 This is a diagram showing changes in the pressure matrix when an adult follows through the access control device in the first embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of a community visitor management device based on AI technology in Embodiment 2 of the present invention; Figure 6 This is a schematic diagram of the structure of a community visitor management system based on AI technology in Example 3 of the present invention.

[0029] Figure numerals: 100, front-end interaction equipment; 200, matrix pressure sensing device; 300, back-end management equipment; 400, access control equipment. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments.

[0031] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0032] This embodiment discloses a community visitor management method and system based on AI technology, which are mainly used to solve the problem of failure of tailgating alarms in outdoor environments in the prior art.

[0033] Embodiment 1 like Figure 1 As shown, this embodiment discloses a community visitor management method based on AI technology, which is used for a community visitor management system with a matrix pressure sensor device installed outdoors. The method of the present invention includes: S10: When the front-end interactive device is used to obtain the first facial image of the visitor, a matrix pressure sensing device is simultaneously used to monitor the pressure matrix of the visitor on the ground in real time.

[0034] S20: Based on the face recognition algorithm, the first face image is matched with the images in the pre-registered face image library. If the match is successful, an opening instruction is sent to the access control device, and the access control opening time is recorded.

[0035] S30: Calculate the pressure gradient difference and modulus value of the pressure matrix after the door access opening time.

[0036] S40: Continue to determine whether any of the following conditions is met: Condition 1: The pressure gradient difference is less than or equal to 0; Condition 2: The pressure gradient difference is greater than 0 and the modulus value is less than the pressure threshold; S50: If all the conditions are not met, a tailgating alarm is issued.

[0037] Through the above technical solutions S10-S50, the method of the present invention uses the time frame after the face recognition to analyze the pressure matrix data of a single person passing. If the pressure gradient difference is less than or equal to 0, it means that a single person is passing or a device with a linearly increasing pressure change is passing. If the pressure gradient difference is greater than 0, it means that the person may be carrying a child or multiple people are passing. At this time, the modulus value of the pressure matrix is ​​introduced to compare with the pressure threshold. If condition two is met, it means that the person is carrying a child and is allowed to pass. Otherwise, it means that multiple people are following and a tailing alarm is issued. Compared with the prior art, the method of the present invention is more suitable for outdoor environments. It uses pressure matrix data to analyze the tailing scenario and is almost not affected by factors such as outdoor direct sunlight, wind-blown tree shadows, smoke and dust, solving the problem of infrared tailing detection tolerance failure in outdoor environments.

[0038] It should be explained that the device whose pressure changes into a linear increase can be a cart, a trolley, a suitcase or other cargo carrier. These devices are all equipped with wheels, and their pressure change characteristics are generally linearly increasing or linearly decreasing. The front-end interactive device in the above S10 mainly includes a display screen and a camera. The camera is used to continuously capture photos of visitors, and multiple first face images are extracted based on multiple photos. Among them, the first face image information obtained from a single photo should include information such as facial contours and three-part and five-eye information. The hardware width and length of the matrix pressure sensing device in the above S10 are adaptively designed according to actual conditions. The monitoring of pressure sensing data and the image monitoring of the front-end interactive device are parallel monitoring processes.

[0039] In order to guide visitors in and out, before S10, the method of the present invention further includes: Identify whether the visitor is in the visitor passage, and if so, execute S10.

[0040] If not, a pop-up window will be sent to the visitor indicating that the service is suspended.

[0041] It needs to be further explained that the above-mentioned suspension service pop-up window is displayed on the display screen of the front-end interactive device. When the visitor wants to go to the non-visitor channel to enter the community, the corresponding channel will refuse service until the visitor goes to the visitor channel to participate in face recognition, thereby achieving weak guidance.

[0042] In this embodiment, the face recognition algorithm in S20 above adopts a multi-frame recognition algorithm. The specific process of matching the first face image with the image in the pre-registered face image library in S20 above is: S21: Calculate the single-frame matching degree between each frame of the first face image and the second face image in the face image library; S22: Calculate the matching mean of multiple single-frame matching degrees; S23: Determine whether the matching mean is higher than the matching threshold, if so, the matching is successful.

[0043] Through the above-mentioned S21-S23 scheme, the method of the present invention continuously captures the facial information of the visitor to obtain a plurality of continuous frames of the first facial image, then calculates the matching degree of the plurality of first facial images and the second facial image, and then adds the plurality of matching degrees to obtain the average value as the final matching average output. This method can improve the accuracy of facial recognition to avoid the problem of insufficient recognition accuracy of a single-frame snapshot. In addition, the above-mentioned scheme can also provide assistance for extending the matrix pressure sensor device to maintain a certain steady-state time.

[0044] Furthermore, if the facial image information matching fails, the method of the present invention further includes: S200: Read the temporary access request from the mobile applicant.

[0045] S201: Verify whether the information of the temporary access request includes the identity ID, access area and contact number. If so, send an opening command to the access control device.

[0046] It should be explained that, in some special cases, a facial matching degree lower than the preset threshold does not necessarily mean that the visitor is a suspicious person. It may also be that the visitor has forgotten to register the facial information in advance. In this case, the above technical solutions S200-S201 can provide visitors with a temporary access application channel to meet the actual needs of the visitors.

[0047] In this embodiment, after the access control device is turned on, it is turned off 3 seconds after the matrix pressure sensor device stops being subjected to force.

[0048] Furthermore, the pressure gradient difference of the pressure matrix after the access control opening time is calculated in the above S30, specifically: (1) In the formula, , The access control opening time. is the pressure gradient difference, For time Pressure Matrix The pressure gradient, For time Pressure Matrix pressure gradient.

[0049] Furthermore, the above time Pressure Matrix The pressure gradient is calculated as: (2) In the formula, Represents the pressure matrix exist The pressure gradient at that moment, Indicated in The pressure matrix of the moment, Indicated in The pressure matrix of the moment, express arrive time interval.

[0050] Furthermore, the calculation formula of the module value in the above S30 is: (3) In the formula, For The modulus of the moment pressure matrix, is the number of rows in the pressure matrix, is the number of columns in the pressure matrix.

[0051] Furthermore, the calculation formula of the pressure threshold in the above S40 is: (4) In the formula, is the pressure threshold, is the number of rows in the pressure matrix, is the number of columns of the pressure matrix, The time when the door is opened The pressure matrix, is the proportionality coefficient.

[0052] It should be noted that, in the process of obtaining the pressure value of the pressure matrix, the pressure values ​​of each matrix unit are different, which is mainly affected by the center of gravity or weight of the human body. In order to eliminate the influence of the center of gravity and weight factors, in this embodiment, Figure 2-Figure 4 As shown, before calculating the pressure gradient difference and modulus value of the pressure matrix, the method of the present invention also includes: if ,Will The value of is replaced by 1. This method of normalizing the pressure values ​​is conducive to data analysis of the change trend of the pressure matrix, thereby improving the accuracy and reliability of the method of the present invention for tailing judgment.

[0053] Combination Figure 2 As shown, it can be seen that when a single person passes through the access control device, the change of his pressure matrix shows a nonlinear sudden drop trend. Therefore, his pressure gradient shows a downward trend as a whole, and the corresponding gradient difference is a negative value or 0 (0 represents a constant decrease or unchanged).

[0054] Specifically, the data monitoring table for the single-pass access control device is: Table 1

[0055] like Figure 3 As shown, when the suitcase is moving forward, its pressure matrix changes in a linearly increasing trend, then becomes constant, and then decreases linearly. The pressure matrix changes generated during its movement meet the above conditions 1 and 2. Specifically, the data monitoring table of the suitcase is: Table 2

[0056] like Figure 4 As shown in the figure, if another adult passes after the face recognition object passes, the pressure matrix will increase nonlinearly, and the pressure gradient difference will suddenly change to a positive value. Specifically, the data monitoring table of adult tailing behavior is as follows: Table 3

[0057] As shown in Table 3, when an adult is following, at the moment when one foot steps into the matrix pressure sensor device, the pressure gradient difference and pressure modulus do not meet conditions one and two, and the following alarm will be triggered.

[0058] Through the above formulas (1)-(4), more accurate data analysis of the pressure matrix changes can be performed to obtain more accurate calculation results. The pressure threshold calculated in formula (4) can be adaptively adjusted according to the face recognition object itself. The pressure threshold is based on the modulus of the recognition object when it stands on its feet, and then multiplied by the pressure coefficient as its threshold. This method can combine the proportional coefficient to adaptively adjust the threshold, which is similar to identifying and limiting the threshold based on the foot size of the recognition object. This method is conducive to improving the accuracy of tailgating detection.

[0059] Preferably, the above proportionality coefficient The value range of is 0.25-0.5. In general, in order to improve the accuracy of the method of the present invention, the above Take 0.5, this ratio can basically be used to distinguish the area of ​​adult soles from that of children. , height The area of ​​a single foot of a child aged 3-6 is , height , height is lower than Children aged 3-6 generally enjoy free tickets or no real-name authentication, and basically do not need facial recognition authentication, so they should be allowed to pass when accompanied by an adult. Considering that the contact area of ​​an adult and a 3-6 year old child with one foot stepping into the matrix pressure sensor device is different, the sole area of ​​a 3-6 year old child is 1.5 times that of an adult. Therefore, based on the above proportional coefficient and the above formula (4), it is possible to distinguish between adult tailgating and child tailgating.

[0060] Embodiment 2 like Figure 5 As shown, this embodiment discloses a community visitor management device based on AI technology, which adopts the community visitor management method based on AI technology recorded in Example 1 to realize visitor face recognition and tailing detection alarm. The device of the present invention includes a front-end interaction device 100, a matrix pressure sensing device 200, a back-end management device 300 and a access control device 400.

[0061] Specifically, the front-end interactive device 100 is composed of a camera and a display screen, which is arranged at the front end of the access control device 400 to obtain the first face image of the visitor. The matrix pressure sensing device 200 is arranged at the front end of the visitor channel to monitor the pressure matrix of the visitor on the ground in real time. The back-end management device 300 is composed of a face recognition module, a pressure gradient difference calculation module and a tailing alarm module. The face recognition module is used to match the first face image with the image in the pre-registered face image library based on the face recognition algorithm. If the match is successful, an opening instruction is issued to the access control device 400 and the access control opening time is recorded. The pressure gradient difference calculation module is used to calculate the pressure gradient difference and modulus value of the pressure matrix after the access control opening time. As for the tailing alarm module, it is used to continuously determine whether any of the following conditions are met: whether the pressure gradient difference is less than or equal to 0; the pressure gradient difference is greater than 0 and the modulus value is less than the pressure threshold; if all are not met, a tailing alarm is issued. The access control device 400 is used to open the gate when receiving the opening instruction.

[0062] In this embodiment, the matrix pressure sensing device 200 can use an 8×8 flexible film pressure sensing sensor, whose model can be M0808M, and whose size can be customized according to actual needs. The periphery of the matrix pressure sensing device 200 can be wrapped with elastic resin to achieve a waterproof protection effect. In addition, the top of the matrix pressure sensing device 200 can be provided with a groove adapted to the shoe upper.

[0063] In other embodiments, the matrix pressure sensing device 200 may be formed by a plurality of pressure sensors (strain gauges) distributed in a matrix.

[0064] It should be noted that the more rows and columns there are in the matrix pressure sensing device 200, the higher the accuracy of the method of the present invention in detecting tailgating. In other embodiments, flexible film pressure sensing sensors with more rows and columns may be used.

[0065] Preferably, the device of the present invention can be applied to application scenarios requiring "one gate, one person" such as childcare centers, kindergartens, medical and nursing communities, or ticket checking platforms.

[0066] Compared with the existing technology, the device of the present invention is equipped with a face recognition algorithm and a pressure tailing detection algorithm, which can accurately identify visitors and issue an alarm for tailing behavior, and has stronger reliability and higher security.

[0067] Embodiment 3 like Figure 6 As shown, this embodiment provides a community visitor management system based on AI technology, including a processor and a memory, the memory storing computer program instructions for implementing a community visitor management method based on AI technology, and when the computer program instructions are executed by the processor, the community visitor management method based on AI technology of the first aspect is implemented.

Claims

1. A community visitor management method based on AI technology, characterized in that: include: When the front-end interactive device (100) is used to obtain a first facial image of a visitor, a matrix pressure sensing device (200) is simultaneously used to monitor in real time the pressure matrix of the visitor on the ground; Based on a face recognition algorithm, the first face image is matched with an image in a pre-registered face image library, and if the match is successful, an opening instruction is sent to the access control device (400), and the access control opening time is recorded; Calculate the pressure gradient difference and modulus of the pressure matrix after the access control opening time; the calculation formula of the pressure gradient difference is: in, , The access control opening time. is the pressure gradient difference, For time Pressure Matrix The pressure gradient, For time Pressure Matrix The pressure gradient; Continuously judging whether any of the following conditions is met: the pressure gradient difference is less than or equal to 0; the pressure gradient difference is greater than 0 and the modulus value is less than a pressure threshold; If none of the conditions are met, a tailing alarm will be issued.

2. The community visitor management method based on AI technology according to claim 1 is characterized in that: exist time Pressure Matrix The pressure gradient is calculated as: In the formula, Represents the pressure matrix exist The pressure gradient at that moment, Indicated in The pressure matrix of the moment, Indicated in The pressure matrix of the moment, express arrive time interval.

3. The community visitor management method based on AI technology according to claim 1 is characterized in that: The calculation formula of the modulus value is: In the formula, For The modulus of the moment pressure matrix, is the number of rows in the pressure matrix, is the number of columns in the pressure matrix.

4. The community visitor management method based on AI technology according to claim 1 is characterized in that: Before calculating the pressure gradient difference and the modulus value of the pressure matrix, the method further includes: like ,Will is replaced with 1.

5. The community visitor management method based on AI technology according to claim 1 is characterized in that: The calculation formula of the pressure threshold is: In the formula, is the pressure threshold, is the number of rows in the pressure matrix, is the number of columns of the pressure matrix, The time when the door is opened The pressure matrix, is the proportionality coefficient.

6. The community visitor management method based on AI technology according to claim 5 is characterized in that: The proportionality factor The value range is 0.25-0.

5.

7. The community visitor management method based on AI technology according to claim 1 is characterized in that: The face recognition algorithm uses a multi-frame recognition algorithm to match the first face image with images in a pre-registered face image library, specifically: Calculate the single frame matching degree between each frame of the first face image and the second face image in the face image library; Calculate the matching mean of multiple single-frame matching degrees; It is determined whether the matching mean is higher than the matching threshold, and if so, the matching is successful.

8. The community visitor management method based on AI technology according to claim 1 is characterized in that: Before using the front-end interactive device (100) to obtain the first facial image of the visitor, the method further comprises: Identify whether the visitor is in the visitor channel. If so, obtain the visitor's first facial image; if not, send a service suspension pop-up window to the visitor.

9. The community visitor management method based on AI technology according to claim 1 is characterized in that: Matching the first face image with images in a pre-registered face image library, if the match fails, the method further includes: Read the temporary access request from the mobile application terminal; Verify whether the information of the temporary access request contains the identity ID, access area and contact number, and if so, send an opening command to the access control device (400).

10. A community visitor management system based on AI technology, characterized in that: It includes a processor and a memory, the memory stores computer program instructions for implementing a community visitor management method based on AI technology, and when the computer program instructions are executed by the processor, the community visitor management method based on AI technology as described in any one of claims 1 to 9 is implemented.

Images