Home entrance guard system

Abstract

The invention relates to a household entrance guard system, and belongs to the field of entrance guard systems. The system comprises a virtual dynamic password module, an access control card recognition module, a semiconductor fingerprint recognition module, a binocular living body face recognition module and an access control risk recognition module, and the access control risk recognition module comprises an abnormity recognition module, an emotion recognition module and a defense module. The anomaly recognition module is used for constructing a facial anomaly recognition model through the facial feature information and outputting facial anomaly recognition information, wherein the facial anomaly recognition information comprises normal information and abnormal information; the emotion recognition module is used for outputting emotion deviation information through feature extraction when the abnormal information is received; and the defense module is used for judging whether to issue an access control releasing instruction or not through the number of persons before access control and face verification when receiving the negative emotion information. According to the invention, the access control security is ensured through the plurality of security identification modules.

Description

Technical Field

[0001] The invention belongs to the technical field of door access control systems, and in particular relates to a household door access control system. Background Art

[0002] Home access control systems play a vital role in modern home security. They can verify individual identity information by using various identification technologies, such as fingerprint recognition, face recognition, and ID card recognition. With the advancement of technology and the development of society, the demand for home security is increasing. As the first line of defense for home security, the importance of access control systems has become increasingly prominent. Home access control systems can effectively prevent unauthorized persons from entering the home, thereby enhancing the sense of security of family members; home access control systems can also effectively protect the privacy of family members. By restricting the entry of unauthorized persons, access control systems can prevent neighbors, strangers, or other irrelevant persons from entering and leaving the home at will, thereby protecting the private lives of family members from being affected; in the event of an emergency, home access control systems can respond quickly and provide necessary security. For example, when family members encounter an emergency, they can quickly call for help through the access control system, or quickly lock the door in an emergency to prevent criminals from entering, effectively protecting the personal safety and property safety of family members.

[0003] However, existing access control systems are easily physically damaged. Once the access control system is damaged from the outside, its security protection function will be lost. This means that even if the system's own recognition technology is advanced, it cannot resist direct physical attacks, which greatly reduces the reliability and security of the entire system. In addition, existing access control systems usually lack the ability to perceive the user's current status. And when the user receives a security threat, such as someone following him, it is impossible to ensure the user's safety through effective means. In this case, even if the user successfully passes the identity authentication, he may fall into a dangerous situation because he cannot convey the danger signal to the system in time. Summary of the invention

[0004] In order to solve the above problems existing in the prior art, the present invention provides a home access control system.

[0005] The purpose of the present invention can be achieved through the following technical solutions: A home access control system, comprising a virtual dynamic password module, an access card recognition module, a semiconductor fingerprint recognition module, a binocular live face recognition module and an access risk recognition module, wherein the access risk recognition module is used to judge the user's emotional fluctuation through facial abnormality recognition information, and to recognize the user's emotional bias information through MediaPipeFaceMesh, CNN network and MTTS-CAN network, so as to judge whether to issue an access release instruction.

[0006] Preferably, the virtual dynamic password module includes an outside door acquisition module and an inside door control module, the outside door acquisition module is used to input a virtual dynamic password, the virtual dynamic password is a combination of a virtual password and a dynamic password, the inside door control module is used to receive the virtual dynamic password, if it contains a preset dynamic password, then send a verification pass instruction to the access control risk identification module.

[0007] Preferably, the access control card identification module is used to identify the access control card with a built-in RFID chip through a card reader. When the access control card is close to the card reader, the electromagnetic wave of the card reader activates the RFID chip and reads the application information. The application information is verified by the access control system host. When the application information is consistent with the registration information entered in the early stage, a verification pass instruction is sent to the access control risk identification module.

[0008] Preferably, the semiconductor fingerprint recognition module is used to recognize fingerprint image data through a capacitive sensor, and when the fingerprint image data matches the preset input fingerprint image data, a verification pass instruction is sent to the access control risk recognition module.

[0009] Preferably, the binocular live face recognition module is used to capture face information through a binocular camera, the face information includes face image information and liveness test information, the liveness test information includes liveness information and non-liveness information, when the face image information is consistent with the preset registered face image and the liveness test information is the liveness information, a verification pass instruction is sent to the access control risk identification module.

[0010] Preferably, the access control risk identification module includes an abnormality identification module, an emotion recognition module and a defense module.

[0011] Preferably, the abnormality recognition module is used to construct a facial abnormality recognition model through facial feature information and output facial abnormality recognition information, and the facial abnormality recognition information includes normal information and abnormal information; the emotion recognition module is used to output emotional bias information through feature extraction when receiving the abnormal information, and the emotional bias information includes positive emotional information, neutral emotional information and negative emotional information; the defense module is used to determine whether to issue a door release command through the number of people in front of the door and face verification when receiving the negative emotional information.

[0012] Preferably, the abnormality identification module specifically includes: When a verification pass instruction is received, facial videos are collected and frame-processed to obtain facial images, and the facial feature information of each facial image is obtained, where the facial feature information includes the degree of eyebrow lift, degree of sclera exposure, distance between the two sides of the nose, and degree of lip opening; The emotional fluctuation reference value of each facial image is calculated based on the facial feature information, and the calculation formula is: , where A is the reference value of emotional fluctuation, α is the degree of eyebrow rise, S is the degree of sclera exposure, and d is the distance between the two sides of the nose. The degree of lip opening; Gathering the emotion fluctuation reference values ​​of each of the facial images to form an emotion fluctuation reference value set, traversing the emotion fluctuation reference value set, screening out the emotion fluctuation maximum value and the emotion fluctuation minimum value, obtaining the emotion fluctuation degree according to the difference between the emotion fluctuation maximum value and the emotion fluctuation minimum value, and constructing the facial abnormality recognition model based on the emotion fluctuation degree; Preset an emotion fluctuation threshold, and obtain the facial abnormality recognition information by comparing the emotion fluctuation with the emotion fluctuation threshold; Determining the emotional fluctuation degree, when the emotional fluctuation degree is greater than or equal to the emotional fluctuation degree threshold, the facial abnormality recognition model outputs the abnormality information and issues an emotional recognition instruction to the access control terminal; When the emotion fluctuation degree is less than the emotion fluctuation degree threshold, the facial abnormality recognition model outputs the normal information and sends a release access control instruction to the access control terminal.

[0013] Preferably, the emotion recognition module specifically includes: When the emotion recognition instruction is received, N facial key points are selected based on the facial image through MediaPipe FaceMesh, key point vectors between adjacent facial key points are obtained, and a surface normal vector set is constructed. The expression of the surface normal vector set is: , where Vn is the surface normal vector set of the nth facial key point, is the key point vector from the nth facial key point to the ith facial key point, is the key point vector from the nth facial key point to the jth facial key point, k is the number of facial key points adjacent to the nth facial key point, and N is the number of facial key points; Extracting a first emotion feature matrix based on the surface normal vector set through a CNN network; Extracting the rPPG signal in the facial video through the MTTS-CAN network and inputting it into the rPPG feature extraction subnetwork to obtain a second emotion feature matrix; fusing the first emotion feature matrix and the second emotion feature matrix to obtain an emotion feature matrix, and obtaining the emotion bias information through a classifier based on the emotion feature matrix; When the positive emotion information and the neutral emotion information are output, a release access control instruction is sent to the access control terminal, and when the negative emotion information is output, a warning instruction is sent to the access control terminal.

[0014] Preferably, the defense module specifically includes: When the alert command is received, the number of people in front of the access control is detected and the binocular live face recognition module is automatically triggered to perform face verification; when the face verification is passed and the number of people in front of the access control is 1, a release access control command is sent to the access control terminal; when the face verification is passed but the number of people in front of the access control is greater than 1, the access control terminal sends a help command to the property end; when the face verification is not passed and the number of people in front of the access control is 1, the access control terminal sends an alarm command to the property end; when the face verification is not passed and the number of people in front of the access control is greater than 1, the access control terminal automatically alarms.

[0015] The beneficial effects of the present invention are: (1) By adopting a virtual dynamic password module, the security of the access control system of the present application has been significantly improved. Compared with the prior art, the present application can effectively prevent the password from being cracked and provide higher security protection when the user is under security threats. The use of a virtual dynamic password makes the password input different each time, which increases the complexity and security of the password, thereby improving the overall security performance of the access control system. By setting up an outside door collection module and an inside door control module, information is collected outside the door and whether the door is opened is controlled inside the door. Even if the access control system is damaged from the outside, the system inside the door can control the door not to open, further ensuring the security of the access control system.

[0016] (2) RFID technology enables contactless identification, avoiding the wear and malfunction problems of traditional contact identification. Secondly, the system can quickly and accurately verify application information, improving the response speed and security of the access control system.

[0017] (3) By introducing a binocular live face recognition module, the problem of face recognition being easily attacked by photos or videos in the existing access control system is solved, further improving the security and reliability of the system. Compared with the existing technology, this application can more effectively prevent illegal intrusion and ensure that users can be protected in a timely manner when they are threatened with security.

[0018] (4) The access control risk identification module first performs abnormal identification to determine whether the person who triggers the access control system has a large emotional fluctuation. If so, it performs emotional discrimination to determine whether the user's emotions are negative, such as panic, tension, etc., so as to determine whether the user's safety is threatened. If the user's safety may be threatened, the number of people captured by the access control system is used to determine whether the user is being followed. If the person is not a registered user, he or she may have bad intentions and preventive measures must be taken. The above measures are closely linked to each other and can maximize the safety of users and venues.

[0019] (5) Obtain emotional fluctuation reference values ​​through facial feature information, taking into account phenomena such as eyebrows rising and increased scleral leakage area that may occur when a person's emotions fluctuate, and comprehensively judge the possible emotional fluctuations that a person may experience.

[0020] (6) Through MediaPipe FaceMesh, CNN network and MTTS-CAN network, the user's emotional bias information is identified to determine whether to issue the access control release command. MediaPipe FaceMesh can detect 468 feature points of the face in real time, which is far beyond the traditional five-point or six-point positioning of the face. Each key point has its corresponding 3D coordinate, which enables MediaPipe to work not only on 2D images, but also to perform simple 3D reconstruction. This high-precision feature point detection capability is particularly important for application scenarios that require accurate facial expression analysis and three-dimensional reconstruction. It is a deep learning model that is particularly suitable for processing image data. The CNN network has significant advantages in network structure, parameter sharing, data input and feature extraction. The MTTS-CAN network is a multi-task time-shifted convolutional attention network with an end-to-end design that uses a time-shift module to perform effective temporal modeling and eliminate various noise sources without additional computational overhead. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0022] Figure 1 The present invention is a flow chart of the steps of a home door entry control system.

[0023] Figure 2 The figure is a schematic diagram of a home access control system. DETAILED DESCRIPTION

[0024] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0025] The working principle and use process of the present invention: See also Figure 1A home access control system includes a virtual dynamic password module, an access card recognition module, a semiconductor fingerprint recognition module, a binocular live face recognition module and an access risk recognition module. The access risk recognition module is used to judge the user's emotional fluctuation through facial abnormality recognition information, and recognize the user's emotional bias information through MediaPipe FaceMesh, CNN network and MTTS-CAN network, so as to judge whether to issue a release access control instruction. The virtual dynamic password module, the access card recognition module, the semiconductor fingerprint recognition module and the binocular live face recognition module can judge whether to issue a verification pass instruction through one of the modules, or can also use multiple modules to judge whether to issue a verification pass instruction at the same time, such as controlling the access control by a dynamic password alone, or by dual encryption of a dynamic password and a semiconductor fingerprint, and the above modules and the access risk recognition module are in a mutually coordinated relationship. After the user passes the verification of the above modules, the verification pass instruction is issued to the access risk recognition module, and the access risk recognition module performs abnormality recognition, emotion recognition and defense.

[0026] In this embodiment, the virtual dynamic password module includes an outside acquisition module and an inside control module, the outside acquisition module is used to input a virtual dynamic password, the virtual dynamic password is a combination of a virtual password and a dynamic password, the inside control module is used to receive the virtual dynamic password, and the preset dynamic password is sent to the access control risk identification module. That is, any number can be added before and after each dynamic password generated, and the verification instruction can be sent when the correct dynamic password is included in the middle. By adopting the virtual dynamic password module, the access control system of the present application has been significantly improved in security. The present application can effectively prevent the password from being cracked and provide higher security protection when the user is threatened with security. The use of virtual dynamic passwords makes each password input different, increases the complexity and security of the password, and thus improves the overall security performance of the access control system. By setting the outside acquisition module and the inside control module, information is collected outside the door, and the door is controlled inside whether to open the door. Even if the access control system is destroyed from the outside, the system inside the door can control the door not to open, further ensuring the security of the access control system.

[0027] In this embodiment, the access card identification module is used to identify the access card with a built-in RFID chip through a card reader. When the access card is close to the card reader, the electromagnetic wave of the card reader activates the RFID chip and reads the application information. The application information is verified by the access control system host. When the application information is consistent with the registration information input in the early stage, a verification instruction is issued to the access control risk identification module. As a preferred embodiment, the card reader can adopt a multi-band design to adapt to different types of RFID chips, further improving the compatibility and recognition efficiency of the system. The access card identification module described in this application has significant advantages over the prior art. First, non-contact identification is achieved through RFID technology, avoiding the wear and failure problems of traditional contact identification. Secondly, the system can quickly and accurately verify the application information, improving the response speed and security of the access control system.

[0028] In this embodiment, the semiconductor fingerprint recognition module is used to identify fingerprint image data through a capacitive sensor. When a finger contacts the capacitive sensor, the distance and discharge speed between the fingerprint ridge and the fingerprint valley and the capacitive sensor are different, thereby forming the fingerprint image data. When the fingerprint image data matches the preset input fingerprint image data, a verification pass instruction is sent to the access control risk recognition module.

[0029] In this embodiment, the binocular live face recognition module is used to capture face information through a binocular camera, the face information includes face image information and liveness test information, the liveness test information includes liveness information and non-liveness information, the binocular camera includes a visible light camera and an infrared light camera, the binocular camera synchronously shoots and uses the triangulation principle to calculate depth information, such as calculating the pupil distance, nose height, distance from eyes to mouth to ears and other spatial information for face recognition positioning, thereby obtaining the face image information, and at the same time judging whether it is a flat photo video or a three-dimensional face according to the depth of each part of the face, and adding liveness detection technology to distinguish whether the recognized object is a real person, thereby obtaining the liveness test information, when the face image information is consistent with the preset registered face image and the liveness test information is the liveness information, a verification pass instruction is issued to the access control risk identification module. This application solves the problem that face recognition in the existing access control system is easily attacked by photos or videos by introducing a binocular live face recognition module, and further improves the security and reliability of the system. Compared with the prior art, the present invention can more effectively prevent illegal intrusions and ensure that users can receive timely protection when they are threatened with security.

[0030] In this embodiment, one or more of the virtual dynamic password module, the access card recognition module, the semiconductor fingerprint recognition module, and the binocular live face recognition module can be triggered simultaneously.

[0031] In this embodiment, the access control risk identification module includes an abnormality identification module, an emotion recognition module and a defense module.

[0032] In this embodiment, the abnormal recognition module is used to construct a facial abnormal recognition model through facial feature information and output facial abnormal recognition information, and the facial abnormal recognition information includes normal information and abnormal information; the emotion recognition module is used to output emotional bias information through feature extraction when receiving the abnormal information, and the emotional bias information includes positive emotional information, neutral emotional information and negative emotional information; the defense module is used to determine whether to issue a release access control instruction through the number of people in front of the access control and face verification when receiving the negative emotional information. This application can identify and handle potential risks in the access control system through the collaborative work of the abnormal recognition module, the emotion recognition module and the defense module of the access control risk recognition module, especially when the user may be threatened with security, to provide effective defense means. Through the abnormal recognition of facial feature information and the extraction of emotional bias information, the user's emotional state can be judged more accurately, and corresponding defense measures can be taken when necessary, thereby improving the security and reliability of the access control system.

[0033] In this embodiment, the abnormality identification module specifically includes: When a verification pass instruction is received, facial videos are collected and frame-processed to obtain facial images, and the facial feature information of each facial image is obtained, wherein the facial feature information includes eyebrow lift, sclera exposure, distance between two nose wings, and lip opening and closing. The eyebrow lift is the angle formed between a line from the starting point of the eyebrow to the end of the eyebrow and a horizontal line, the sclera exposure is the area of ​​sclera leakage, the distance between two nose wings is the distance between the lowest ends of two nose wings, and the lip opening and closing is the area of ​​lip opening; The emotional fluctuation reference value of each facial image is calculated based on the facial feature information, and the calculation formula is: , where A is the reference value of emotional fluctuation, α is the degree of eyebrow rise, S is the degree of sclera exposure, and d is the distance between the two sides of the nose. The degree of lip opening; Gathering the emotion fluctuation reference values ​​of each of the facial images to form an emotion fluctuation reference value set, traversing the emotion fluctuation reference value set, screening out the emotion fluctuation maximum value and the emotion fluctuation minimum value, obtaining the emotion fluctuation degree according to the difference between the emotion fluctuation maximum value and the emotion fluctuation minimum value, and constructing the facial abnormality recognition model based on the emotion fluctuation degree; Preset an emotion fluctuation threshold, and obtain the facial abnormality recognition information by comparing the emotion fluctuation with the emotion fluctuation threshold; Determining the emotional fluctuation degree, when the emotional fluctuation degree is greater than or equal to the emotional fluctuation degree threshold, the facial abnormality recognition model outputs the abnormality information and issues an emotional recognition instruction to the access control terminal; When the emotion fluctuation degree is less than the emotion fluctuation degree threshold, the facial abnormality recognition model outputs the normal information and sends a release access control instruction to the access control terminal.

[0034] In this embodiment, the emotion recognition module specifically includes: When the emotion recognition instruction is received, N facial key points are selected based on the facial image through MediaPipe FaceMesh, key point vectors between adjacent facial key points are obtained, and a surface normal vector set is constructed. The expression of the surface normal vector set is: , where Vn is the surface normal vector set of the nth facial key point, is the key point vector from the nth facial key point to the ith facial key point, is the key point vector from the nth facial key point to the jth facial key point, k is the number of facial key points adjacent to the nth facial key point, and N is the number of facial key points; A first emotion feature matrix is ​​extracted through a CNN network based on the surface normal vector set. Specifically, the CNN network extracts the first emotion feature matrix as follows: the CNN network extracts features of input information through multiple convolutional layers, and each convolutional layer uses multiple convolutional kernels to scan the input image to extract local features, which include edge, texture and other information; after being processed by an activation function, the feature map is downsampled through a pooling layer; after stacking multiple convolutional and pooling layers, the CNN network extracts higher-level features layer by layer; in a fully connected layer, the CNN network converts the extracted high-dimensional features into emotion feature vectors, and integrates the emotion feature vectors to obtain the first emotion feature matrix; Extracting the rPPG signal in the facial video through the MTTS-CAN network and inputting it into the rPPG feature extraction subnetwork to obtain a second emotion feature matrix; The rPPG feature extraction subnetwork specifically comprises: performing standardization and detrending processing on the rPPG signal and removing high-frequency noise and low-frequency noise, extracting the main frequency components through fast Fourier transform and forming a one-dimensional vector, inputting the one-dimensional vector into a one-dimensional CNN network to obtain multiple feature vectors, and integrating the feature vectors to obtain the second emotion feature matrix; fusing the first emotion feature matrix and the second emotion feature matrix to obtain an emotion feature matrix, and obtaining the emotion bias information through a classifier based on the emotion feature matrix; When the positive emotional information and the neutral emotional information are output, a command to release the access control is issued to the access control terminal, and when the negative emotional information is output, a warning command is issued to the access control terminal. MediaPipe FaceMesh can detect 468 feature points of the face in real time, a capability far beyond the traditional five-point or six-point positioning of the face. Each key point has its corresponding 3D coordinate, which enables MediaPipe to work not only on 2D images, but also to perform simple 3D reconstruction. This high-precision feature point detection capability is particularly important for application scenarios that require accurate facial expression analysis and three-dimensional reconstruction; it is a deep learning model that is particularly suitable for processing image data; the CNN network has significant advantages in network structure, parameter sharing, data input, and feature extraction; the MTTS-CAN network is a multi-task time-shifted convolutional attention network that adopts an end-to-end design and uses a time-shifting module to perform effective time modeling and eliminate various noise sources without additional computational overhead.

[0035] In this embodiment, the defense module specifically includes: When the alert command is received, the number of people in front of the access control is detected and the binocular live face recognition module is automatically triggered to perform face verification. When the face verification is passed, that is, the person in front of the access control is a registered person, and the number of people in front of the access control is 1, a release access control command is issued to the access control terminal. When the face verification is passed but the number of people in front of the access control is greater than 1, the access control terminal sends a help command to the property end. When the face verification is not passed and the number of people in front of the access control is 1, the access control terminal sends an alarm command to the property end. When the face verification is not passed and the number of people in front of the access control is greater than 1, the access control terminal automatically alarms.

[0036] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device.

[0037] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0038] The program code included on the computer readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF, etc., or any suitable combination of the above. The computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or their combinations, and the programming language includes object-oriented programming languages-such as Java, Smalltalk, C++, and also includes conventional procedural programming languages-such as "C" language or similar programming languages. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0039] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A home access control system, characterized in that: The home access control system includes a virtual dynamic password module, an access card recognition module, a semiconductor fingerprint recognition module, a binocular live face recognition module and an access risk recognition module. The access risk recognition module is used to judge the user's emotional fluctuation through facial abnormality recognition information, and recognize the user's emotional bias information through MediaPipe FaceMesh, CNN network and MTTS-CAN network, so as to judge whether to issue a release access control instruction. The virtual dynamic password module includes an outside door acquisition module and an inside door control module. The outside door acquisition module is used to input a virtual dynamic password, which is a combination of a virtual password and a dynamic password. The inside door control module is used to receive the virtual dynamic password, and if it contains a preset dynamic password, a verification instruction is issued to the access risk recognition module. The access card recognition module is used to identify an access card with a built-in RFID chip through a card reader. When the access card is close to the card reader, the electromagnetic wave of the card reader activates the RFID chip and reads the application information. The application information is verified by the access control system host. When the application information is consistent with the previously input When the registered information entered is consistent, a verification pass instruction is sent to the access control risk identification module. The semiconductor fingerprint identification module is used to identify fingerprint image data through a capacitive sensor. When the fingerprint image data is consistent with the preset input fingerprint image data, a verification pass instruction is sent to the access control risk identification module. The binocular live face recognition module is used to capture face information through a binocular camera. The face information includes face image information and liveness test information. The liveness test information includes liveness information and non-liveness information. When the face image information is consistent with the preset registered face image and the liveness test information is the liveness information, a verification pass instruction is sent to the access control risk identification module. The access control risk identification module includes an abnormality recognition module, an emotion recognition module and a defense module.

2. The home access control system according to claim 1, characterized in that: The abnormality recognition module is used to construct a facial abnormality recognition model through facial feature information and output facial abnormality recognition information, and the facial abnormality recognition information includes normal information and abnormal information; the emotion recognition module is used to output emotional bias information through feature extraction when receiving the abnormal information, and the emotional bias information includes positive emotional information, neutral emotional information and negative emotional information; the defense module is used to determine whether to issue a door release command based on the number of people in front of the door and face verification when receiving the negative emotional information.

3. The home access control system according to claim 2, characterized in that: The abnormality identification module specifically includes: When a verification pass instruction is received, facial videos are collected and frame-processed to obtain facial images, and the facial feature information of each facial image is obtained, where the facial feature information includes the degree of eyebrow lift, degree of sclera exposure, distance between the two sides of the nose, and degree of lip opening; The emotional fluctuation reference value of each facial image is calculated based on the facial feature information, and the calculation formula is: , where A is the reference value of emotional fluctuation, α is the degree of eyebrow rise, S is the degree of sclera exposure, and d is the distance between the two sides of the nose. The degree of lip opening; Gathering the emotion fluctuation reference values ​​of each of the facial images to form an emotion fluctuation reference value set, traversing the emotion fluctuation reference value set, screening out the emotion fluctuation maximum value and the emotion fluctuation minimum value, obtaining the emotion fluctuation degree according to the difference between the emotion fluctuation maximum value and the emotion fluctuation minimum value, and constructing the facial abnormality recognition model based on the emotion fluctuation degree; Preset an emotion fluctuation threshold, and obtain the facial abnormality recognition information by comparing the emotion fluctuation with the emotion fluctuation threshold; Determining the emotional fluctuation degree, when the emotional fluctuation degree is greater than or equal to the emotional fluctuation degree threshold, the facial abnormality recognition model outputs the abnormality information and issues an emotional recognition instruction to the access control terminal; When the emotion fluctuation degree is less than the emotion fluctuation degree threshold, the facial abnormality recognition model outputs the normal information and sends a release access control instruction to the access control terminal.

4. The home access control system according to claim 3, characterized in that: The emotion recognition module specifically includes: When the emotion recognition instruction is received, N facial key points are selected based on the facial image through MediaPipe FaceMesh, key point vectors between adjacent facial key points are obtained, and a surface normal vector set is constructed. The expression of the surface normal vector set is: , where Vn is the surface normal vector set of the nth facial key point, is the key point vector from the nth facial key point to the ith facial key point, is the key point vector from the nth facial key point to the jth facial key point, k is the number of facial key points adjacent to the nth facial key point, and N is the number of facial key points; Extracting a first emotion feature matrix based on the surface normal vector set through a CNN network; Extracting the rPPG signal in the facial video through the MTTS-CAN network and inputting it into the rPPG feature extraction subnetwork to obtain a second emotion feature matrix; fusing the first emotion feature matrix and the second emotion feature matrix to obtain an emotion feature matrix, and obtaining the emotion bias information through a classifier based on the emotion feature matrix; When the positive emotion information and the neutral emotion information are output, a release access control instruction is sent to the access control terminal, and when the negative emotion information is output, a warning instruction is sent to the access control terminal.

5. The home access control system according to claim 4, characterized in that: The defense module specifically includes: When the alert command is received, the number of people in front of the access control is detected and the binocular live face recognition module is automatically triggered to perform face verification; when the face verification is passed and the number of people in front of the access control is 1, a release access control command is sent to the access control terminal; when the face verification is passed but the number of people in front of the access control is greater than 1, the access control terminal sends a help command to the property end; when the face verification is not passed and the number of people in front of the access control is 1, the access control terminal sends an alarm command to the property end; when the face verification is not passed and the number of people in front of the access control is greater than 1, the access control terminal automatically alarms.

Images