A reliable user identity authentication method based on RFID

Abstract

The present invention provides a user identity authentication method based on RFID trust, which includes two modules: user registration and user identification. The present invention uses semantic addresses and physical addresses to complete the tracing and positioning of RFID tags in the system, and constructs an intrusion detection method of a two-level defense system. Without increasing the cost of tags, the present invention effectively completes the real-time monitoring of the RFID system and the identification of tag cloning attacks.

Description

Technical Field

[0001] The present invention belongs to the field of information security. Specifically, the present invention relates to a reliable user identity authentication method based on RFID. Background Art

[0002] With the development of computer and Internet of Things technology, non-manual user identity authentication technologies emerge in an endless stream. In various places where access control of personnel is required, such as residential community access control, company access control, government workplace, factory warehouse, campus access control, etc., the user identity authentication process is an indispensable step. In the prior art, methods and systems that can be used for user identity authentication include: face recognition authentication, smart cards that mark the user's personal identity;

[0003] However, existing user identity authentication methods and systems have certain risks of privacy leakage and great risks of being deceived and attacked, especially in some confidential places such as company confidential archives, R&D laboratories, high-end community access control and home entry, etc., which urgently need high-reliability user identity authentication scenarios. Holding smart information such as the contactless IC card of the campus card system lacks authentication. The card holder can obtain access rights through card authentication, and there is a lack of authentication of the identity of the card owner itself, which leads to a high risk of being deceived and attacked. Moreover, current face recognition authentication is mostly based on face image recognition. This recognition method itself has a great risk of privacy leakage. For example, the user's facial information is exposed to the camera and can be easily captured remotely, resulting in information leakage or even theft. In addition, the risk of being deceived and attacked based on face image recognition is also relatively high. Although there are face recognition authentication methods based on facial geometry on the market, this authentication method reduces background interference to a certain extent and increases the difficulty of deception attacks, but it is still easy to be hacked. For example, 3D printed faces can be used for deception attacks, and when biometric recognition technology is converted into data and stored, attackers can even use this data to track users. When users pass by the camera, records will be left in the system, and this is done without the user's awareness, that is, in an unconscious state. Summary of the invention

[0004] Technical problem: In view of the privacy leakage and deception attack problems existing in the prior art, the purpose of the present invention is to provide a reliable user identity authentication method based on RFID, which can complete the user identity authentication process and realize card theft prevention by accurately identifying the user's facial features and efficiently matching the biometric features read by the electronic tag inside the user's card. Compared with the prior art, the user identity authentication method provided by the present invention improves the card theft prevention level on the one hand and improves the reliability of facial recognition on the other hand.

[0005] Technical solution: The present invention provides a reliable user identity authentication method based on RFID. The authentication method is implemented based on the registration of authorized users. After successful registration, users can pass the identification authentication by holding a card, such as Figure 1 As shown, the RFID-trusted user identity authentication method includes the following steps:

[0006] User registration process: The user's facial biometric feature T is obtained from the client by using facial PLP recognition and uploaded to the server. The server extracts the user's biometric feature vector from the facial biometric feature T as biometric feature information M and generates a digital signature MD; the digital signature MD is encrypted using the RSA public key algorithm, where the user's public key is stored in the user's contactless IC;

[0007] User identification process: If a non-user is unconsciously identified in a specific area, the system will not directly identify their facial biometrics, considering that the contactless IC is not in the specific identification area. Only when the user enters the identification area of ​​the RFID reader with the contactless IC, the digital signature MD is decrypted using the public key in the contactless IC and the corresponding private key in the server to obtain the user's biometric information M. At the same time, facial PLP recognition is started, and when it is determined that there is only one face in the identification area, the user's facial biometric T' is obtained, and the user's generated feature vector is extracted from the facial biometric as the biometric information M', and the biometric information M' is matched with the biometric information M. If the match is successful, the user authentication is passed.

[0008] It should be noted that the facial PLP recognition method is based on the facial privacy protection anti-spoofing (Privacy leak prevention) system, hereinafter referred to as PLP. This system is based on the extraction of two COTSRFID tag array-based three-dimensional geometric and internal biomaterial features of the face for identification. These features are not only difficult to obtain but also difficult to forge, so they can effectively prevent attacks. The hardware system of PLP mainly includes RFID tags, readers and antennas. The extraction of facial biometric features through RSS is essentially the propagation and processing of radio frequency signals. The propagation of radio frequency signals in the system is divided into three stages: from antenna to face, reflection to face and from face to antenna; COTSRFID equipment is used to implement facial PLP recognition. Users only need to stay in front of the tag array for a few seconds to complete identity authentication. Facial PLP recognition has an authentication success rate of more than 95.7% and an identical error rate of about 44%. It is a more effective and secure identification method; in addition, the PLP method is effective in combating three mainstream deception attacks, including 2D, 2D +It has obvious advantages over 3D spoofing attacks, and even 3D mask attacks can be effectively prevented. Therefore, this recognition method will effectively improve the security and accuracy of face recognition of the present invention.

[0009] The propagation of the radio frequency signal in the facial PLP recognition of the present invention is divided into three stages: from the antenna to the face, reflected to the face, and from the face to the antenna.

[0010] ①In the first stage of propagation, the value of R is defined by the power P, which is proportional to the square of the amplitude a. R can be expressed as:

[0011]

[0012] Where D is a constant and the amplitude A usually causes an exponential loss per unit of propagation distance, which can be expressed as e -a .

[0013] ②In the second propagation stage, the RF signal reaches the surface of the face. The arriving signal can be divided into two parts: the directly reflected part and the part refracted into the face. Since the internal structure of the face is composed of a variety of biological materials, it is assumed that each person's face is a unique mixed material composed of multiple layers of materials. Because the RF signal refracted to the face must pass through multiple layers of tissue and undergo multiple reflections before escaping the internal tissue of the face, its power will decay exponentially, and eventually only very low-power signals will escape from the inside of the face, so this part of the signal can be ignored. Therefore, the focus of this part is to analyze the power loss caused by reflection on the face surface. The power ratio of the signal before and after reflection on the face surface is called the power reflection coefficient, so the corresponding amplitude is calculated by the following formula:

[0014]

[0015] Among them, A a and A b is the amplitude of the reflected signal before and after. PER It is related to the mixed material of the face and the incident angle, so by obtaining R PER The facial features of the face can be obtained. This method effectively avoids the behavior of forging faces and using face images to deceive.

[0016] ③ The third propagation stage, the face signal reaches the antenna. Combining the first and second stages of propagation, the amplitude of the received signal can be obtained, and according to the solution formula of R, the R′ of the tag receiving signal can be calculated.

[0017]

[0018] Substituting the data for calculation, we can find that the value of R is related to the distance d between the person and the label array. The greater the change in the distance of the label array, the smaller the change in the value of R. In order to better eliminate the influence of factors unrelated to the face structure on the experimental results, we combine the label array and make a difference elimination factor d on the R of the two labels.

[0019] After eliminating the factors, it is concluded that the difference between the R values ​​of the tags can reveal the three-dimensional geometric shape and internal biomaterial characteristics of the face, that is, the difference in facial features of different users can be verified through the coupling of RFID tags and wireless readers. This method effectively completes the user's identity authentication and ensures the privacy of the user's facial recognition.

[0020] Due to factors such as unnatural facial movement and hardware imperfections during signal propagation, the phase value collected from the RFID reader will include noise signals, which will reduce the accuracy and security of the entire recognition process. Therefore, it is further preferred that the facial PLP recognition of the present invention also includes a signal denoising and relocation process;

[0021] The denoising process specifically includes: eliminating noise by expanding continuous phase values, and filtering out abnormal phase values ​​in the window by setting a sample window of size 5, thereby obtaining the average value of other normal values, thereby eliminating noise.

[0022] The redirection process specifically includes: for subsequent facial feature extraction, R and phase values ​​are rearranged according to the layout of the marker array, thereby obtaining a new sequence.

[0023] Based on the face PLP recognition model optimized by the denoising and relocalization process, the phase change process of the three stages of signal propagation is as follows:

[0024] The RF signal (θ b ) before entering the face can be expressed as:

[0025]

[0026] Where θ0 is the initial phase value of the transmitter, and d is the distance between the tag array and the chin. In this patent, it is considered that once the signal enters the face, it cannot escape. Therefore, this patent only considers the emission occurring on the surface of the human face. In this case, the RF signal (θ a )The phase after emission is:

[0027] θ a =θ b +Δθ

[0028] Where Δθ represents the phase change caused by reflection on the face, and Δθ is also related to facial features. in ) is greater than the Brewster angle (β B ), a phase change occurs:

[0029]

[0030] Therefore, θ p1 It is determined by β1 and ε1, that is, the phase change value (θ p1 ) can reveal the internal biological material and three-dimensional facial geometry of the human face. Combined with the phase change in the third stage, the final phase of the received RF signal is:

[0031]

[0032] Where θ t2 is the additional phase shift caused by the tag characteristics. Similarly, in order to remove the factor d, the present invention uses θ j Subtract θ i , the result can be expressed as:

[0033]

[0034] It is worth noting that since the phase of the received RF signal is a periodic function with a period of 2π, there is a 2kπ term in theory. Similar to the R analysis, the present invention can conclude that the phase difference between the tags (θ i -θ j ) can represent the three-dimensional geometry and internal biomaterial features of the human face.

[0035] The present invention generates a feature vector by extracting the user's biometric features. This feature vector is used as a digital signature to achieve confidentiality and non-repudiation. The background recognition system is assumed to be a trusted party, which has a high ability to prevent spoofing attacks compared to traditional communication parties. The RSA public key algorithm is selected for encryption. Preferably, during the user registration process, the step of generating a digital signature MD from the biometric information M is as follows: the extracted biometric information M is encrypted using its own private key and a digital signature MD is generated, which is formalized as follows:

[0036] MD=E U_id (M)

[0037] Under normal user use, the system uses the corresponding card identifier as the card's private key. The attacker cannot obtain the identifier U_id specified by the system, so theoretically, the attacker cannot obtain the user's private key. U_id indicates that the user's biometric information M is encrypted using its own private key and a digital signature is generated.

[0038] The contactless IC card corresponds to the identification in the system, which serves as the private key for user communication; ps_num represents the number assigned to the card by the system, and the corresponding formula is as follows:

[0039] U_id=Hash(ps_num)

[0040] The user's digital signature is MD. The extracted facial information plaintext is encrypted with its own private key. After the system receives the corresponding ciphertext, it can use the card's public key to decrypt it accordingly. The card's user name is used as the public key U. When the system receives the corresponding ciphertext during subsequent recognition, it can use the card's public key to decrypt it accordingly. The card's user name is used as the public key U.

[0041] The encrypted transmission information specifically uses the digital signature and user information together with the public key of the system to generate the ciphertext C;

[0042] C=E P_id (MD||M)=E P_id (E U_id (M)||M)

[0043] Among them, P_id represents the system identification, which is the public key of the system; P is the private key of the system, which is specified by the adopted communication protocol. When the system decrypts, it can use its own private key and the user's public key. Compared with the traditional digital signature, the user's private key is obtained by the one-way function Hash value, and the private key is automatically generated by the system and the corresponding card on the KPI platform;

[0044] M'=MD||M=D P (C)

[0045] The system determines the user's identity: The system uses the private key specified by PKI to decrypt the ciphertext, obtain the user's plaintext and digital signature, and can verify the integrity of the information through the digital signature.

[0046] Beneficial effects:

[0047] The authentication method proposed in the present invention uses the face recognition component as a biometric identifier, works closely with the biometric key manager, and combines with the identity authentication server to provide the identity authentication result to determine the use of the card, eliminating the singleness of obtaining the card and using it immediately. At the same time, the human biometrics have long-term stability. In the process of intelligent biotechnology application, the biometrics have long-term stability, coupled with technical rationality, the interference of external social factors can be almost completely eliminated;

[0048] The authentication method proposed in the present invention involves that the user's facial information will not directly expose the facial biometric features on the display during the RFID identification process, and the information will not be intercepted by a third party during the verification process. In the future, the development program will be combined with the NFC wireless function to realize the identification data exchange and identity authentication of compatible devices within a short distance without a card;

[0049] The present invention is based on RFID technology combined with modern cryptography, which ensures the real-name nature of the RFID card and can protect user safety to a certain extent. The technical application range of the present invention is wide, and it is widely used in the fields of RFID card anti-theft, company access control, community access control, etc.

[0050] The present invention proposes an intrusion detection method based on facial recognition, which uses semantic addresses and physical addresses to complete the tracing and positioning of RFID tags in the system, and constructs a two-level defense system. Without increasing the cost of tags, it effectively completes the real-time monitoring of the RFID system and the identification of tag cloning attacks. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is the overall flow chart of the system;

[0052] Figure 2 A flowchart for privacy protection;

[0053] Figure 3 A flowchart for capturing information flow. DETAILED DESCRIPTION

[0054] 1. System Structure

[0055] The present invention is mainly divided into the following three parts in structure: public key encryption privacy protection, anti-spoofing attack information capture and identity authentication.

[0056] PKI public key infrastructure is required for digital signature. PKI public key infrastructure is a system or platform that provides public key encryption and digital signature services for the purpose of managing keys and certificates. An organization can establish a secure network environment by adopting the PKI framework to manage keys and certificates.

[0057] PKI mainly consists of four parts: X.509 format certificates and certificate revocation list CRL (X.509V2), CA operation protocol, CA management protocol, and CA policy formulation. Using zero-knowledge proof security protocol, better confidentiality can be achieved through negotiation with KPI.

[0058] The present invention uses a facial privacy protection anti-spoofing system that extracts two COTS RFID tag array-based facial 3D geometry and internal biomaterial features for identification. These features are not only difficult to obtain but also difficult to forge, thus effectively preventing attacks.

[0059] The hardware system of PLP mainly includes RFID tags, readers and antennas. The extraction of facial features through RSS mainly includes the propagation of RF signals and the extraction of facial features. The propagation of RF signals in the system is divided into three stages: from antenna to face, reflected to face and from face to antenna.

[0060] In the specific implementation of the present invention, facial PLP recognition is implemented using COTSRFID equipment. The user only needs to stay in front of the tag array for a few seconds to complete the identity verification. Facial PLP recognition has an authentication success rate of more than 95.7% and an identical error rate of about 44%. It is a more effective and secure identification method. In addition, the PLP method has obvious advantages in combating three mainstream spoofing attacks, including 2D, 2D+ and 3D spoofing attacks, and even 3D mask attacks can be effectively prevented. Therefore, this recognition method will effectively improve the security and accuracy of face recognition of the present invention.

[0061] 2. Methodology

[0062] 1. Privacy protection

[0063] like Figure 2 As shown in the figure, for the prediction stage of the online server-side face recognition privacy protection model, the server will require the client's RFID reading device to collect a photo of the face to be recognized. After the server receives the private face data transmitted by the client, it uses the face recognition privacy protection model to predict the face in conjunction with the face data to be recognized. The recognition result is still in the form of secret shared ciphertext, and the recognition result is returned to the client. The client reassembles the secret to obtain the plaintext recognition result, completing the face recognition.

[0064] 2. Information Capture

[0065] like Figure 3 As shown, based on the above-mentioned privacy protection method based on image recognition, the system can record the user's facial information in real time through the RFID reader and tag array, and no longer use the traditional camera to obtain image information, and perform algorithm modeling on the obtained RFID information; detect and identify the number of faces in the model. When it is recognized that the number of faces in the model established at a certain moment is one, the system will use the face alignment algorithm to directly extract and verify the facial information; when the number of faces is greater than one, the information status is set as to be analyzed, the appearance time of the model to be analyzed is recorded, and the facial area corresponding to all the faces in the model to be analyzed is detected and recorded, and the facial model corresponding to the face is located and rotated and aligned. The facial features obtained after processing are identified using a deep residual network to extract the facial features of all the faces in the face model.

[0066] The areas of the faces identified in the model to be analyzed are compared, and the actual distances between all faces and the RFID reader are estimated based on the comparison results; a three-dimensional face model is constructed for the facial feature points of all faces using the face alignment algorithm PRNet, and the constructed three-dimensional face model is rotated to obtain the face rotation angle, and the deflection angles of all faces on the model and the plane where the RFID reader is located are obtained based on the face rotation angle; the duration of face appearance is determined based on the appearance time of the model to be analyzed and the time of deframing the video stream information; based on the duration, it is determined whether the deflection angle and the actual distance exceed the corresponding set thresholds within a predetermined time; if two faces reach the threshold at the same time, the system issues a warning and re-identifies the face.

[0067] 3. Information matching – identity verification

[0068] GB / T36651-2018In the application scenario, fingerprint recognition, face recognition and other components act as biometric identifiers, work closely with the biometric key manager running in the trusted environment of the smartphone, and interact with the identity authentication server of the identity service provider to provide the identity authentication result for the relying party (i.e., the application service provider). The system realizes the identity authentication of users based on the session protocol between the biometric key manager and the identity authentication server.

[0069] The present invention provides formal definitions and efficient security techniques for turning noisy information into a key that can be used in any cryptographic application, in particular for reliably and securely verifying biometric data. The present invention's techniques are applicable not only to biometric information, but to any key material. Unlike traditional cryptographic keys, these keys have two characteristics: they cannot be exactly copied, and they cannot be uniformly distributed. The present invention proposes two primitives: a fuzzy extractor reliably extracts near-uniform randomness R from its input; the extractor is fault-tolerant, i.e., R will be the same even if the input changes, as long as it remains reasonably close to the original. Therefore, R can be used as a key in cryptographic applications. A secure sketch produces public information about its input w that does not reveal w, but allows accurate recovery of w given another value close to w. Therefore, it can be used to reliably reproduce error-prone biometric inputs without the security risks inherent in storing them. The primitives defined by the present invention are formally secure and general, generalizing much previous work. Furthermore, the present invention provides nearly optimal construction of both primitives for various "closeness" measures of the input data, such as Hamming distance, edit distance, and set divergence.

[0070] (1) Bilinear pairing

[0071] Linear mapping defines three prime p-order group multiplication cyclic groups G1 and G2, and defines a mapping relationship on these three groups. Composite order bilinear pairing can use the orthogonality of subgroups to complete the security proof under the premise of realizing more complex functions. The modified Weil pairing and Tate pairing are acceptable mappings. According to the basic definition of the above bilinear pairing (G1, G2, e), given (P, αP, α 2 P,…,α n P), where α∈Z G The security of the scheme depends on the difficulty of dealing with the following problems.

[0072] Problem 1 is to calculate e(P,P) 1 / α , the second problem is to find a pair (c,1 / (a+c)P)∈Z G G1, where c∈Z G .

[0073] (2) Fuzzy Extractor Technology

[0074] The main process of the fuzzy extractor technique is to extract a unique ID from the biometric sample b and allow a tolerable threshold. This means that if the fuzzy extractor technique is used on biometric samples A and a from the same user, the same identity information ID will be obtained, but the sample distance between the two biometric samples A and a must satisfy dis(A,a)≤1. Specifically, the definition of the fuzzy extractor is given by {M,l,t} and two algorithms {Gen,Rep}, where M is the metric space with biometric sample points, l is the length of the output identity string ID, and t is the error threshold.

[0075] The two algorithms {Gen, Rep} are described as follows:

[0076] ①Gen: is a probabilistic extraction algorithm that takes biometric data: A∈M as input and outputs identity information ID and public recovery parameters PAR.

[0077] ②Rep: is a deterministic retrieval algorithm that takes another biometric data a∈M and PAR as input, and outputs identity information ID if dis(A,a)≤t; if dis(A,a)>t, outputs the termination symbol ⊥. Among them, (Gen) and (Rep) have the following characteristics.

[0078] ① The generator Gen outputs an extracted string R∈{0,1} and an auxiliary string P∈{0,1} when input w∈M.

[0079] ② The reproduction program Rep takes an element w∈M and a bit string P∈{0,1}. As input. The correctness property of the fuzzy extractor guarantees that if dis(w,w0)≤t, and R, P are (R,P)←Gen(w), then Rep(w0,P)=R. If dis(w,w0)>t, then the output cannot be guaranteed.

[0080] ③ The security property ensures that for any distribution W on M with minimum entropy m, the string R is nearly uniform even for someone who observes P.

[0081] It is nearly uniform even for someone observing P: if (R,P)←Gen(w), then SD((R,P),(U,P))≤∈. Fuzzy extractors allow one to extract some randomness R from w, and then successfully reproduce R from any string w using the auxiliary string P produced during the initial extraction.

[0082] The nearly uniform random bits output by the fuzzy extractor can be used in any cryptographic context where uniform random bits are required (e.g., for keys), with slight non-uniformity of the bits potentially reducing security by no more than their distance from uniformity.

[0083] (3) Matching threshold setting

[0084] Considering that the existing facial recognition system has a unified constant for each user's judgment threshold, there is a phenomenon that some faces are recognized slowly. The present invention sets different thresholds according to different users. The method of setting the threshold is as follows: when registering, the system prompts the user to enter facial information 3 to 5 times, and records the multiple information entered by the user in the database. Through the information entered when the user registers, the system performs normalized comparison calculations based on the feature differences of multiple groups of data and the feature differences of all users in the database, and sets the critical value that can complete the match during the user identification process. The database script is called to update the database in real time to ensure that during the verification process, the verification time can be dynamically minimized based on the overall information of all users.

Claims

1. A user identity authentication method based on RFID trust, characterized in that: The method includes two modules: user registration and user identification; User registration process: The user's facial biometrics T is obtained from the client by using facial PLP recognition and uploaded to the server. The server extracts the user's biometric vector from the facial biometrics as biometric information M and generates a digital signature MD; the digital signature MD is encrypted using the RSA public key algorithm, where the user's public key is stored in the user's contactless IC; User identification process: When a user enters the identification area of ​​the RFID reader with a contactless IC, the public key in the contactless IC and the corresponding private key in the server are used to decrypt the digital signature MD to obtain the user's biometric information T. At the same time, facial PLP recognition is started, and when it is determined that there is only one face in the identification area, the user's facial biometric feature T' is obtained, and the user's generated feature vector is extracted from the facial biometric feature as the biometric information M', and the biometric information M' is matched with the biometric information M. If the match is successful, the user authentication is passed; The step of generating a digital signature and encrypting the biometric information specifically includes: encrypting the extracted biometric information M using its own private key and generating a digital signature MD, which is formalized as follows: MD=E U_id (M) Among them, U_id represents the user's biometric information M, which is encrypted with its own private key and generates a digital signature; the contactless IC card corresponds to the identifier in the system, which serves as the private key for user communication; ps_num represents the number assigned to the card by the system, and the corresponding formula is as follows: U_id=Hash(ps_num) The user's digital signature is MD. The extracted facial information plaintext is encrypted with its own private key. After the system receives the corresponding ciphertext, it can be decrypted using the card's public key. The card's user name is used as the public key U. After receiving the corresponding ciphertext in subsequent recognition, the card's public key can be used to decrypt it. The card's user name is used as the public key U. The encrypted transmission information is specifically to use the digital signature and user information together with the system's public key to generate the ciphertext C. C=E P_id (MD||M)=E P_id (E U_id (M)||M) Among them, P_id represents the system identification, which is the public key of the system; P is the private key of the system, which is specified by the adopted communication protocol; when the system performs decryption, it uses its own private key and the user's public key; compared with the traditional digital signature, the user's private key is obtained by the one-way function Hash value, and the private key is automatically generated by the system and the corresponding card on the KPI platform; M'=MD||M=D P (C); The facial PLP recognition includes the propagation of radio frequency signals and signal processing, wherein the signal processing includes the denoising and relocation process of the radio frequency signals; Based on the face PLP recognition model optimized by the denoising and relocalization process, the phase change process of the three stages of signal propagation is as follows: The RF signal transmitted by the antenna is b The phase before entering the face is expressed as: Where θ0 is the initial phase value of the transmitter, and d is the distance between the tag array and the chin. Once the signal enters the face, it cannot escape. Only the transmission occurring on the surface of the human face is considered to calculate the RF signal θ a The phase after emission is: i a =θ b +Δθ Where Δθ represents the phase change caused by reflection on the face, and Δθ is related to facial features; after the RF signal is reflected, when the incident angle β in Greater than Brewster angle β B When , a phase change occurs: Therefore, θ p1 Determined by β1 and ε1, that is, the phase change value θ caused by the face reflection p1 , representing the internal biological material and three-dimensional facial geometric features of the human face; combined with the phase change in the third stage, the final phase of the received RF signal is: Where θ t2 is the additional phase shift caused by the label characteristics. Similarly, in order to remove the factor d, we use θ j Subtract θ i , the result is expressed as: Since the phase of the received RF signal is a periodic function with a period of 2π, the phase difference between the tags is θ i -θ j Represents the three-dimensional geometry and biomaterial internal features of the human face.

2. The user identity authentication method based on RFID trustworthiness according to claim 1 is characterized in that: The denoising process specifically includes: eliminating noise by expanding continuous phase values, and filtering out abnormal phase values ​​in the window by setting a sample window of size 5, thereby obtaining the average value of other normal values, thereby eliminating noise.

3. The user identity authentication method based on RFID trustworthiness according to claim 1 is characterized in that: The relocation process specifically includes: for subsequent facial feature extraction, R and phase values ​​are rearranged according to the layout of the marker array to obtain a new sequence.

4. The user identity authentication method based on RFID trustworthiness according to claim 1, characterized in that: The facial PLP recognition sets different thresholds according to different users, and the threshold setting method is as follows: when registering, the user is prompted to enter facial information 3 to 5 times, and the multiple information entered by the user is recorded in the database; based on the information entered by the user when registering, the system performs normalized comparison calculations based on the feature differences of multiple groups of data and the feature differences of all users in the database, and sets the critical value for completing the match during the user identification process.

Images