Campus security payment system based on biological feature recognition

Abstract

The invention discloses a campus security payment system based on biological feature recognition, which relates to the technical field of biological feature recognition and comprises a campus security payment data acquisition module, a biological feature extraction module, an identity verification module, an error analysis module, a payment verification module and a campus security payment module. The identity verification module is used for constructing an identity verification model according to a feature extraction result of the biological recognition data and then carrying out identity verification on campus users, and the payment verification module is used for constructing a payment verification model according to a face recognition error rate, a fingerprint input error rate, a palmprint input error rate and payment data and then carrying out payment verification on the campus users. The data acquisition technology, the biological feature extraction technology and the error analysis technology in the system are closely combined with the modern information technology, the analysis of the identity verification coefficient and the payment verification coefficient is realized, and the intelligent degree in the campus security payment process based on biological feature recognition is remarkably enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of biometric identification, and particularly relates to a campus security payment system based on biometric identification. Background Art

[0002] In today's digital age, the pursuit of security and convenience in campus management is constantly increasing. A campus security payment system based on biometric identification has emerged as the times require. Traditional campus payments rely on campus cards, and there are problems such as easy loss, being stolen and swiped, and cumbersome reissuance procedures. At the same time, the efficiency of manual identity verification is low, and it is easy to cause congestion in places with large traffic such as canteens and supermarkets. With the rapid development of biometric identification technology, technologies such as fingerprint, palmprint, and face recognition are becoming increasingly mature, providing an opportunity for the innovation of the campus payment system. These technologies utilize the characteristics of human biometric features, can accurately identify identities, effectively reduce the disadvantages of traditional payments, and apply them to the campus payment scenario, which can achieve fast and secure payments and improve the informatization level of campus management, greatly improving the payment efficiency and laying a solid foundation for building a smart and secure campus;

[0003] Although there have been great progress in the direction of biometric identification in the existing technology, there are still some problems to be optimized. It is difficult for the existing technology to perform identity verification and payment verification for campus users' security payments through multiple types of biometric identification methods, which may cause problems such as inaccurate biometric identification and insecure payment environments. Summary of the Invention

[0004] To achieve the above objectives, the present invention is realized through the following technical solutions: A campus security payment system based on biometric identification, including a campus security payment data collection module, a biometric feature extraction module, an identity verification module, an error analysis module, a payment verification module, and a campus security payment module, wherein each module is communicatively connected;

[0005] The campus security payment data collection module collects campus security payment data including biometric identification data, biometric verification data, and payment data, and preprocesses the collected data, providing a data basis for the implementation of subsequent module functions;

[0006] The biometric feature extraction module is used to extract features from the preprocessed biometric identification data and biometric verification data, providing data support for the construction of subsequent models;

[0007] The identity verification module constructs an identity verification model through the feature extraction results of biometric identification data, and then verifies the identities of campus users;

[0008] The error analysis module is used to analyze the face recognition error rate, fingerprint input error rate, and palmprint input error rate;

[0009] The payment verification module constructs a payment verification model by combining the error rates of face recognition, fingerprint entry, palmprint entry, and payment data, and then conducts payment verification for campus users. It comprehensively considers the possibilities that not all faces may be fully entered or other faces may be mis-entered during face recognition, and the impact of residual biological traces on the security of payment verification on fingerprint and palmprint scanners during fingerprint and palmprint entry, thereby improving the security level of campus payment.

[0010] The campus secure payment module verifies the secure payment of campus users based on the identity verification result and the payment verification result, and then realizes the secure payment of campus users.

[0011] A further improvement of the technical solution of the present invention lies in that: the error analysis module is composed of a face recognition error analysis unit, a fingerprint entry error analysis unit, and a palmprint entry error analysis unit. The functions of each unit are as follows:

[0012] The face recognition error analysis unit obtains the face entry rate, the mutation rate of face texture grayscale values, and the abnormal face recognition rate through the feature extraction results of biometric data, and then analyzes the face recognition error rate.

[0013] The fingerprint entry error analysis unit analyzes the fingerprint capture error rate based on the feature extraction results of biometric verification data.

[0014] The palmprint entry error analysis unit analyzes the palmprint entry error rate by using the feature extraction results of biometric verification data.

[0015] A further improvement of the technical solution of the present invention lies in that: the process of collecting campus secure payment data by the campus secure payment data collection module includes:

[0016] Combining different types of collection devices and data entry technologies to collect biometric data, biometric verification data, and payment data of campus users. The collection devices include fingerprint scanners, cameras, and palmprint scanners.

[0017] Specifically, the fingerprints, face images, and palmprints of campus users are collected through fingerprint scanners, cameras, and palmprint scanners respectively; cameras are installed in the vertical direction of the fingerprint scanner and the palmprint scanner to obtain the vertical images of the fingerprint scanner and the palmprint scanner when acquiring campus user information, providing a data basis for subsequent error analysis of the residual biological information on the fingerprint scanner and the palmprint scanner; the payment amount is obtained through data entry.

[0018] The biometric data includes fingerprints, face images, and palmprints; the biometric verification data includes the vertical images of the fingerprint scanner and the palmprint scanner; the payment data is the payment amount.

[0019] Perform data cleaning and data standardization on the collected payment data, perform noise removal and grayscale processing on the collected vertical images of face images, fingerprint scanners, and palm scanners, perform contrast adjustment and grayscale processing on the collected fingerprints and palm prints, set time stamps for the preprocessed biometric data, biometric verification data, and payment data, and adjust the set time stamps to achieve synchronization of the collection times of the biometric data, biometric verification data, and payment data.

[0020] A further improvement of the technical solution of the present invention lies in that: in the biometric feature extraction module, the process of extracting features from the preprocessed biometric data and biometric verification data includes:

[0021] Combining the Python language and the OpenCV software library, extract face feature recognition data, fingerprint feature recognition data, and palm print feature recognition data from the fingerprints, face images, and palm prints of campus users respectively;

[0022] The face feature recognition data includes facial feature indices, facial contour indices, skin texture grayscale values, and facial feature texture grayscale values; the fingerprint feature recognition data includes fingerprint shape indices, ridge density, number of bifurcation points, and number of isolated points; the palm print feature recognition data includes the number and length of main lines in the palm print, the angle between main lines, and wrinkle density;

[0023] Using Photoshop software, extract the fingerprint verification image grayscale value and the palm print verification image grayscale value from the vertical images of the fingerprint scanner and the palm scanner respectively;

[0024] Integrate the face feature recognition data, fingerprint feature recognition data, palm print feature recognition data, fingerprint verification image grayscale value, palm print verification image grayscale value, and the preprocessed payment data to generate a biometric verification data set, and divide the biometric verification data set into a training set and a test set, and the ratio of the training set to the test set is 8:2.

[0025] A further improvement of the technical solution of the present invention lies in that: in the identity verification module, the process of constructing an identity verification model and then performing identity verification on campus users includes:

[0026] Extract the face feature recognition data, fingerprint feature recognition data, and palm print feature recognition data from the biometric verification data set;

[0027] Combining the training set data with the multiple linear regression algorithm, using the face feature recognition data, fingerprint feature recognition data, and palm print feature recognition data as inputs and the identity verification coefficient as the output, learn the linear relationship between the face feature recognition data, fingerprint feature recognition data, palm print feature recognition data, and the identity verification coefficient, and train the identity verification model;

[0028] Input the test set data into the authentication model, adjust the intercept term and regression coefficients of the authentication model to optimize the performance of the authentication model, deploy the optimized authentication model to the system, and combine the face feature recognition data, fingerprint feature recognition data, and palmprint feature recognition data to output the corresponding authentication coefficient;

[0029] The expression of the authentication model is:

[0030] SF = α0 + α1z1 + α2z2 + α3z3 + α4z4 + α5s1 + α6s2 + α7s3 + α8s4 + α9x1 + α 10 x2 + α 11 x3 + α 12 x4 + θ

[0031] Among them, SF is the authentication coefficient, and α1, α2, α3, α4, α5, α6, α7, α8, α9, α 10 , α 11 and α 12 are the regression coefficients of the five - sense organ index, facial contour index, skin texture gray value, five - sense organ texture gray value, fingerprint shape index, line density, number of bifurcation points, number of isolated points, number of main lines and their lengths in the palmprint, included angle between main lines, and wrinkle density respectively; z1, z2, z3, and z4 are the five - sense organ index, facial contour index, skin texture gray value, and five - sense organ texture gray value respectively; s1, s2, s3, and s4 are the fingerprint shape index, line density, number of bifurcation points, and number of isolated points respectively; x1, x2, x3, and x4 are the number of main lines and their lengths, included angle between main lines, and wrinkle density in the palmprint respectively, and α0 and θ are the intercept term and error term of the authentication model.

[0032] Authenticate the campus users based on the authentication coefficient to obtain the authentication result, and the authentication result includes authentication passed and authentication failed;

[0033] Specifically, when the authentication coefficient is lower than 0.5, it indicates that the corresponding campus user fails the authentication; when the authentication coefficient is higher than 0.5, it indicates that the corresponding campus user passes the authentication.

[0034] A further improvement of the technical solution of the present invention is that: in the face recognition error analysis unit, the analysis process of the face recognition error rate includes:

[0035] Normalize the face feature recognition data, and calculate the average values of the five - sense organ index, facial contour index, skin texture gray value, and five - sense organ texture gray value respectively;

[0036] Weights are assigned to the average values of the calculated facial feature index, facial contour index, skin texture grayscale value, and facial feature texture grayscale value. The face entry rate is obtained through the weights of the average value of the facial feature index and the average value of the facial contour index; the mutation rate of the face texture grayscale value is obtained through the weights of the average value of the skin texture grayscale value and the average value of the facial feature texture grayscale value.

[0037] Thresholds for the facial feature index, facial contour index, and facial feature texture grayscale value are set respectively. Based on the set thresholds, the abnormal face recognition rate is obtained through calculation.

[0038] Weights are assigned to the obtained face entry rate, mutation rate of the face texture grayscale value, and abnormal face recognition rate respectively. Using the weighted average method, the face recognition error rate is obtained through analysis and integrated into the biometric verification dataset.

[0039] Specifically, the processes of obtaining the face entry rate, mutation rate of the face texture grayscale value, abnormal face recognition rate, and face recognition error rate are as follows:

[0040]

[0041]

[0042] Among them, R1, R2, R3, and RT are the face entry rate, mutation rate of the face texture grayscale value, abnormal face recognition rate, and face recognition error rate respectively; a1, a2, a3, a4, w1, w2, and w3 are the weights of the average value of the facial feature index, average value of the facial contour index, average value of the skin texture grayscale value, average value of the facial feature texture grayscale value, face entry rate, mutation rate of the face texture grayscale value, and abnormal face recognition rate respectively. and are the average values of the facial feature index, facial contour index, skin texture grayscale value, and facial feature texture grayscale value respectively; z1, z2, and z4 are the facial feature index, facial contour index, and facial feature texture grayscale value respectively; Z1, Z2, and Z4 are the thresholds of the facial feature index, facial contour index, and facial feature texture grayscale value respectively.

[0043] A further improvement of the technical solution of the present invention lies in: for the fingerprint entry error analysis unit, the analysis process of the fingerprint entry error rate includes:

[0044] Extract the grayscale values of the fingerprint verification images in the biometric verification dataset. Using the training set data and combining with the linear regression algorithm, taking the grayscale values of the fingerprint verification images as the input and the fingerprint entry error rate as the output, learn the linear relationship between the grayscale values of the fingerprint verification images and the fingerprint entry error rate, and train the fingerprint entry error analysis model.

[0045] Input the test set data into the fingerprint entry error analysis model, adjust the parameters of the fingerprint entry error analysis model, optimize the fingerprint entry error analysis model, obtain the final fingerprint entry error analysis model, combine the fingerprint verification image gray value, output the corresponding fingerprint entry error rate, and integrate the fingerprint entry error rate into the biometric verification data set;

[0046] The expression of this fingerprint entry error analysis model is:

[0047] Q = β0 + β1q1 + τ

[0048] Among them, Q is the fingerprint entry error rate, β1 is the regression coefficient of the fingerprint verification image gray value, q1 is the fingerprint verification image gray value, and β0 and τ are the intercept term and error term of the fingerprint entry error analysis model.

[0049] A further improvement of the technical solution of the present invention lies in: for the palmprint entry error analysis unit, the analysis process of the palmprint entry error rate includes:

[0050] Extract the palmprint verification image gray value in the biometric verification data set, use the training set data, combine the linear regression algorithm, take the palmprint verification image gray value as the input, take the palmprint entry error rate as the output, learn the linear relationship between the palmprint verification image gray value and the palmprint entry error rate, and train the palmprint entry error analysis model;

[0051] Input the test set data into the palmprint entry error analysis model, adjust the parameters of the palmprint entry error analysis model, optimize the palmprint entry error analysis model, obtain the final palmprint entry error analysis model, combine the palmprint verification image gray value, output the corresponding palmprint entry error rate, and integrate the palmprint entry error rate into the biometric verification data set;

[0052] The expression of this palmprint entry error analysis model is as follows:

[0053] P = γ0 + γ1p1 + ε

[0054] Among them, P is the palmprint entry error rate, p1 is the palmprint verification image gray value, γ1 is the regression coefficient of the palmprint verification image gray value, and γ0 and ε are the intercept term and error term of the palmprint entry error analysis model respectively.

[0055] A further improvement of the technical solution of the present invention lies in: for the payment verification module, the process of constructing a payment verification model and then performing payment verification on campus users includes:

[0056] Extract the payment data, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate in the biometric verification data set;

[0057] Using the training set data and combining with the multiple linear regression algorithm, taking payment data, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate as inputs and the payment verification coefficient as the output, learning the linear relationship between payment data, face recognition error rate, fingerprint entry error rate, palmprint entry error rate, and payment verification data, and training the payment verification model;

[0058] Input the test set data into the payment verification model, adjust the intercept term and regression coefficients of the payment verification model, optimize the payment verification model, deploy the optimized payment verification model to the system, and combine with payment data, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate to output the corresponding payment verification coefficient;

[0059] The expression of the payment verification model is as follows:

[0060] ZF = ρ0 + ρ1U + ρ2RT + ρ3Q + ρ4P + σ

[0061] Where, ZF is the payment verification coefficient, ρ1, ρ2, ρ3, and ρ4 are the regression coefficients of payment data, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate respectively, U, RT, Q, and P are the payment amount, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate respectively, and ρ0 and σ are the intercept term and error term of the payment verification model respectively;

[0062] Based on the payment verification coefficient, conduct payment verification on campus users to obtain the payment verification result, and the payment verification result includes payment verification passed and payment verification failed;

[0063] Specifically, when the payment verification coefficient is lower than 0.5, it indicates that the corresponding campus user's payment verification fails; when the payment verification coefficient is higher than 0.5, it indicates that the corresponding campus user's payment verification passes.

[0064] A further improvement of the technical solution of the present invention lies in: for the campus security payment module, the implementation process of the secure payment of campus users includes:

[0065] S1. Set the identity verification as the first verification process and the payment verification as the second verification process;

[0066] S2. Based on the identity verification result, when the identity verification of the campus user fails, do not perform the payment verification and switch to the biometric recognition mode to re-perform the identity verification; when the identity verification of the campus user passes, perform the payment verification, where the biometric recognition mode includes face image recognition, fingerprint recognition, and palmprint recognition;

[0067] S3. Based on the payment verification result, when the payment verification of the campus user fails, the secure payment operation is not performed; when the payment verification of the campus user passes, the secure payment operation is performed.

[0068] The beneficial effects of the present invention are as follows: In a campus secure payment system based on biometric recognition according to the present invention, compared with the traditional campus secure payment system based on biometric recognition, the data acquisition technology, biometric extraction technology, error analysis technology and modern information technology in the system of the present invention are closely combined to accurately capture the biometric recognition data, biometric verification data and payment data of campus users. Through feature extraction, face feature recognition data, fingerprint feature recognition data, palmprint feature recognition data, the gray value of the fingerprint verification image and the gray value of the palmprint verification image are obtained, and then the analysis of the identity verification coefficient and the payment verification coefficient is realized, solving the problem that it is difficult to perform identity verification and payment verification for campus user secure payment through multiple types of biometric recognition methods in the prior art, which may cause inaccurate biometric recognition and an insecure payment environment, ensuring that the method in the present invention can refine the dynamic monitoring standard for a campus secure payment system based on biometric recognition within a more accurate range, making the monitored data a more accurate indicator under the same conditions. The research and application of this method significantly enhance the degree of intelligence in the campus secure payment process based on biometric recognition. Description of the Drawings

[0069] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0070] Figure 1 It is a block diagram of a campus secure payment system based on biometric recognition according to the present invention;

[0071] Figure 2 It is a flowchart of campus secure payment verification. Detailed Embodiments

[0072] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0073] Such as Figure 1 、 Figure 2As shown in the figure, the present invention provides a campus security payment system based on biometric recognition, including a campus security payment data collection module, a biometric feature extraction module, an identity authentication module, an error analysis module, a payment verification module, and a campus security payment module. Among them, each module is communicatively connected;

[0074] The campus security payment data collection module collects campus security payment data including biometric recognition data, biometric verification data, and payment data, and preprocesses the collected data, providing a data basis for the implementation of subsequent module functions;

[0075] The biometric feature extraction module is used to extract features from the preprocessed biometric recognition data and biometric verification data, providing data support for the construction of subsequent models;

[0076] The identity authentication module constructs an identity authentication model through the feature extraction results of biometric recognition data, and then authenticates the identity of campus users;

[0077] The error analysis module is used to analyze the face recognition error rate, fingerprint input error rate, and palmprint input error rate;

[0078] The payment verification module combines the face recognition error rate, fingerprint input error rate, palmprint input error rate, and payment data to construct a payment verification model, and then verifies the payment of campus users. It comprehensively considers the possibilities that the entire face may not be fully input and other faces may be misinput during face recognition, as well as the impact of residual biological traces on the fingerprint scanner and palmprint scanner on the security of payment verification during fingerprint input and palmprint input, improving the security level of campus payment;

[0079] The campus security payment module verifies the secure payment of campus users based on the identity authentication result and the payment verification result, and then realizes the secure payment of campus users.

[0080] Among them, the error analysis module consists of a face recognition error analysis unit, a fingerprint input error analysis unit, and a palmprint input error analysis unit. The functions of each unit are as follows:

[0081] The face recognition error analysis unit obtains the face entry rate, the mutation rate of face texture gray value, and the abnormal face recognition rate through the feature extraction results of biometric recognition data, and then analyzes the face recognition error rate;

[0082] The fingerprint input error analysis unit analyzes the fingerprint input error rate based on the feature extraction results of biometric verification data;

[0083] The palmprint input error analysis unit analyzes the palmprint input error rate using the feature extraction results of biometric verification data.

[0084] Campus Security Payment Data Collection Module. The process of collecting campus security payment data includes:

[0085] Combining different types of collection devices and data entry technologies to collect biometric data, biometric verification data, and payment data of campus users. Among them, the collection devices include fingerprint scanners, cameras, and palm scanners;

[0086] Specifically, collect the fingerprints, face images, and palm prints of campus users through fingerprint scanners, cameras, and palm scanners respectively; install cameras in the vertical direction of the fingerprint scanner and palm scanner to obtain the vertical images of the fingerprint scanner and palm scanner when obtaining campus user information, providing a data basis for subsequent error analysis of the residual biometric information on the fingerprint scanner and palm scanner; obtain the payment amount through data entry;

[0087] Biometric data includes fingerprints, face images, and palm prints; biometric verification data includes the vertical images of fingerprint scanners and palm scanners; payment data is the payment amount;

[0088] Perform data cleaning and data standardization processing on the collected payment data, perform noise removal and grayscale processing on the collected face images, vertical images of fingerprint scanners and palm scanners, perform contrast adjustment and grayscale processing on the collected fingerprints and palm prints, set time stamps for the preprocessed biometric data, biometric verification data, and payment data, and adjust the set time stamps to achieve the synchronization of the collection times of biometric data, biometric verification data, and payment data.

[0089] Biometric Feature Extraction Module. The process of extracting features from the preprocessed biometric data and biometric verification data includes:

[0090] Combining the Python language and the OpenCV software library, respectively extract face feature recognition data, fingerprint feature recognition data, and palm print feature recognition data from the fingerprints, face images, and palm prints of campus users;

[0091] Face feature recognition data includes facial feature indices, facial contour indices, skin texture grayscale values, and facial feature texture grayscale values; fingerprint feature recognition data includes fingerprint shape indices, ridge density, number of bifurcation points, and number of isolated points; palm print feature recognition data includes the number and length of main lines in the palm print, the angle between main lines, and wrinkle density;

[0092] Use Photoshop software to respectively extract the fingerprint verification image grayscale value and the palm print verification image grayscale value from the vertical images of the fingerprint scanner and palm scanner;

[0093] Integrate face feature recognition data, fingerprint feature recognition data, palmprint feature recognition data, fingerprint verification image grayscale values, palmprint verification image grayscale values, and preprocessed payment data to generate a biometric verification dataset, and divide the biometric verification dataset into a training set and a test set, with the ratio of the training set to the test set being 8:2.

[0094] The process of the identity verification module constructing an identity verification model and then verifying the identity of campus users includes:

[0095] Extract the face feature recognition data, fingerprint feature recognition data, and palmprint feature recognition data from the biometric verification dataset;

[0096] Combine the training set data with the multiple linear regression algorithm, use the face feature recognition data, fingerprint feature recognition data, and palmprint feature recognition data as inputs, and the identity verification coefficient as the output, learn the linear relationship between the face feature recognition data, fingerprint feature recognition data, palmprint feature recognition data, and the identity verification coefficient, and train the identity verification model;

[0097] Input the test set data into the identity verification model, adjust the intercept term and regression coefficients of the identity verification model, optimize the performance of the identity verification model, deploy the optimized identity verification model to the system, and combine the face feature recognition data, fingerprint feature recognition data, and palmprint feature recognition data to output the corresponding identity verification coefficient;

[0098] The expression of this identity verification model is:

[0099] SF = α0 + α1z1 + α2z2 + α3z3 + α4z4 + α5s1 + α6s2 + α7s3 + α8s4 + α9x1 + α 10 x2 + α 11 x3 + α 12 x4 + θ

[0100] Among them, SF is the identity verification coefficient, α1, α2, α3, α4, α5, α6, α7, α8, α9, α 10 , α 11 and α 12They are the regression coefficients of the facial feature indexes, facial contour indexes, skin texture gray values, facial feature texture gray values, fingerprint shape indexes, ridge density, number of bifurcation points, number of isolated points, number of main lines and their lengths in the palmprint, included angle between main lines, and wrinkle density, respectively; z1, z2, z3, and z4 are the facial feature indexes, facial contour indexes, skin texture gray values, and facial feature texture gray values, respectively; s1, s2, s3, and s4 are the fingerprint shape indexes, ridge density, number of bifurcation points, and number of isolated points, respectively; x1, x2, x3, and x4 are the number of main lines and their lengths in the palmprint, included angle between main lines, and wrinkle density, respectively, and α0 and θ are the intercept term and error term of the identity verification model;

[0101] Based on the identity verification coefficient, identity verification is performed on campus users to obtain the identity verification result, where the identity verification result includes passing the identity verification and failing the identity verification;

[0102] Specifically, when the identity verification coefficient is lower than 0.5, it indicates that the corresponding campus user fails the identity verification; when the identity verification coefficient is higher than 0.5, it indicates that the corresponding campus user passes the identity verification.

[0103] The face recognition error analysis unit, and the analysis process of the face recognition error rate includes:

[0104] Normalize the face feature recognition data, and calculate the average values of the facial feature indexes, facial contour indexes, skin texture gray values, and facial feature texture gray values, respectively;

[0105] Assign weights to the average values of the calculated facial feature indexes, facial contour indexes, skin texture gray values, and facial feature texture gray values, and obtain the face entry rate through the average value weights of the facial feature indexes and facial contour indexes; obtain the face texture gray value mutation rate through the average value weights of the skin texture gray values and facial feature texture gray values;

[0106] Set the thresholds of the facial feature indexes, facial contour indexes, and facial feature texture gray values respectively, and obtain the abnormal face recognition rate through calculation based on the set thresholds;

[0107] Assign weights to the obtained face entry rate, face texture gray value mutation rate, and abnormal face recognition rate respectively, adopt the weighted average method, analyze and obtain the face recognition error rate, and integrate the face recognition error rate into the biometric verification dataset;

[0108] Specifically, the obtaining processes of the face entry rate, face texture gray value mutation rate, abnormal face recognition rate, and face recognition error rate are as follows:

[0109]

[0110] Among them, R1, R2, R3 and RT are the face entry rate, the mutation rate of the face texture gray value, the abnormal face recognition rate and the face recognition error rate respectively; a1, a2, a3, a4, w1, w2 and w3 are the average value of the facial feature index, the average value of the facial contour index, the average value of the skin texture gray value, the average value of the facial feature texture gray value, the face entry rate, the mutation rate of the face texture gray value and the weight of the abnormal face recognition rate respectively; and are the average values of the facial feature index, the facial contour index, the skin texture gray value and the facial feature texture gray value respectively; z1, z2 and z4 are the facial feature index, the facial contour index and the facial feature texture gray value respectively; Z1, Z2 and Z4 are the thresholds of the facial feature index, the facial contour index and the facial feature texture gray value respectively.

[0111] Fingerprint entry error analysis unit. The analysis process of the fingerprint entry error rate includes:

[0112] Extract the gray value of the fingerprint verification image in the biometric verification dataset. Using the training set data and combining with the linear regression algorithm, take the gray value of the fingerprint verification image as the input and the fingerprint entry error rate as the output, learn the linear relationship between the gray value of the fingerprint verification image and the fingerprint entry error rate, and train the fingerprint entry error analysis model;

[0113] Input the test set data into the fingerprint entry error analysis model, adjust the parameters of the fingerprint entry error analysis model, optimize the fingerprint entry error analysis model, obtain the final fingerprint entry error analysis model, combine with the gray value of the fingerprint verification image, output the corresponding fingerprint entry error rate, and integrate the fingerprint entry error rate into the biometric verification dataset;

[0114] The expression of this fingerprint entry error analysis model is:

[0115] Q = β0 + β1q1 + τ

[0116] Among them, Q is the fingerprint entry error rate, β1 is the regression coefficient of the gray value of the fingerprint verification image, q1 is the gray value of the fingerprint verification image, and β0 and τ are the intercept term and the error term of the fingerprint entry error analysis model respectively.

[0117] Palmprint entry error analysis unit. The analysis process of the palmprint entry error rate includes:

[0118] Extract the gray value of the palmprint verification image in the biometric verification dataset. Using the training set data and combining with the linear regression algorithm, take the gray value of the palmprint verification image as the input and the palmprint entry error rate as the output, learn the linear relationship between the gray value of the palmprint verification image and the palmprint entry error rate, and train the palmprint entry error analysis model;

[0119] Input the test set data into the palmprint entry error analysis model, adjust the parameters of the palmprint entry error analysis model, optimize the palmprint entry error analysis model, obtain the final palmprint entry error analysis model, combine the grayscale values of the palmprint verification images, output the corresponding palmprint entry error rate, and integrate the palmprint entry error rate into the biometric verification dataset;

[0120] The expression of the palmprint entry error analysis model is as follows:

[0121] P = γ0 + γ1p1 + ε

[0122] Where P is the palmprint entry error rate, p1 is the grayscale value of the palmprint verification image, γ1 is the regression coefficient of the grayscale value of the palmprint verification image, and γ0 and ε are the intercept term and error term of the palmprint entry error analysis model respectively.

[0123] The payment verification module, constructing a payment verification model, and the process of performing payment verification on campus users includes:

[0124] Extract the payment data, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate from the biometric verification dataset;

[0125] Using the training set data, combined with the multiple linear regression algorithm, taking the payment data, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate as inputs and the payment verification coefficient as the output, learn the linear relationship between the payment data, face recognition error rate, fingerprint entry error rate, palmprint entry error rate, and payment verification data, and train the payment verification model;

[0126] Input the test set data into the payment verification model, adjust the intercept term and regression coefficient of the payment verification model, optimize the payment verification model, deploy the optimized payment verification model to the system, and combine the payment data, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate to output the corresponding payment verification coefficient;

[0127] The expression of the payment verification model is as follows:

[0128] ZF = ρ0 + ρ1U + ρ2RT + ρ3Q + ρ4P + σ

[0129] Where ZF is the payment verification coefficient, ρ1, ρ2, ρ3, and ρ4 are the regression coefficients of the payment data, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate respectively, U, RT, Q, and P are the payment amount, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate respectively, and ρ0 and σ are the intercept term and error term of the payment verification model respectively;

[0130] Perform payment verification on campus users based on the payment verification coefficient to obtain the payment verification result, where the payment verification result includes passing the payment verification and failing the payment verification;

[0131] Specifically, when the payment verification coefficient is lower than 0.5, it indicates that the corresponding campus user fails the payment verification; when the payment verification coefficient is higher than 0.5, it indicates that the corresponding campus user passes the payment verification.

[0132] For the campus security payment module, the implementation process of the secure payment of campus users includes:

[0133] S1. Set the identity verification as the first verification process and the payment verification as the second verification process;

[0134] S2. Based on the identity verification result, when the campus user fails the identity verification, do not perform the payment verification and switch to the biometric recognition mode to re-perform the identity verification; when the campus user passes the identity verification, perform the payment verification, where the biometric recognition mode includes face image recognition, fingerprint recognition, and palmprint recognition;

[0135] S3. Based on the payment verification result, when the campus user fails the payment verification, do not perform the secure payment operation; when the campus user passes the payment verification, perform the secure payment operation.

[0136] First, collect biometric data, biometric verification data, and payment data by combining different types of collection devices and data entry technologies, and preprocess the collected data; secondly, extract features from the preprocessed biometric data and biometric verification data to obtain face feature recognition data, fingerprint feature recognition data, palmprint feature recognition data, fingerprint verification image gray values, and palmprint verification image gray values; then, construct an identity verification model through the face feature recognition data, fingerprint feature recognition data, and palmprint feature recognition data, and then perform identity verification on campus users; then, analyze the face recognition error rate, fingerprint entry error rate, and palmprint entry error rate in turn, and then combine the payment data to construct a payment verification model to perform payment verification on campus users; finally, based on the identity verification result and the payment verification result, verify the secure payment of campus users, and then realize the secure payment of campus users.

[0137] As described above, it is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. A campus security payment system based on biometric recognition, including a campus security payment data collection module, a biometric extraction module, an identity authentication module, an error analysis module, a payment verification module, and a campus security payment module, where, Each module is communicatively connected, characterized in that: The campus security payment data collection module collects campus security payment data including biometric data, biometric verification data, and payment data, and preprocesses the collected data; The biometric feature extraction module is used to extract features from the preprocessed biometric data and biometric verification data; The identity verification module constructs an identity verification model through the feature extraction results of biometric data, and then verifies the identity of campus users; The error analysis module is used to analyze the face recognition error rate, fingerprint entry error rate, and palmprint entry error rate; The payment verification module constructs a payment verification model by combining the face recognition error rate, fingerprint entry error rate, palmprint entry error rate, and payment data, and then verifies the payment of campus users; The campus security payment module verifies the secure payment of campus users based on the identity verification result and the payment verification result, and then realizes the secure payment of campus users.

2. The campus security payment system based on biometric recognition according to claim 1, characterized in that: The error analysis module consists of a face recognition error analysis unit, a fingerprint entry error analysis unit, and a palmprint entry error analysis unit. The functions of each unit are as follows: The face recognition error analysis unit obtains the face entry rate, the mutation rate of face texture gray values, and the abnormal face recognition rate through the feature extraction results of biometric data, and then analyzes the face recognition error rate; The fingerprint entry error analysis unit analyzes the fingerprint entry error rate based on the feature extraction results of biometric verification data; The palmprint entry error analysis unit analyzes the palmprint entry error rate by using the feature extraction results of biometric verification data.

3. The campus security payment system based on biometric recognition according to claim 2, characterized in that: The campus security payment data collection module. The collection process of campus security payment data includes: Combining different types of collection devices and data entry technologies to collect biometric data, biometric verification data, and payment data of campus users. The collection devices include fingerprint scanners, cameras, and palmprint scanners; The biometric data includes face images, fingerprints, and palmprints; the biometric verification data includes the vertical images of fingerprint scanners and palmprint scanners; the payment data is the payment amount; Perform data cleaning and data standardization processing on the collected payment data, perform noise removal and grayscale processing on the collected face images, vertical images of fingerprint scanners and palmprint scanners, perform contrast adjustment and grayscale processing on the collected fingerprints and palmprints, set time stamps for the preprocessed biometric data, biometric verification data, and payment data, and adjust the set time stamps to achieve the synchronization of the collection times of biometric data, biometric verification data, and payment data.

4. The campus security payment system based on biometric recognition according to claim 3, characterized in that: The biometric feature extraction module. The process of extracting features from the preprocessed biometric data and biometric verification data includes: Combining the Python language and the OpenCV software library, respectively extract face feature recognition data, fingerprint feature recognition data, and palmprint feature recognition data from the fingerprints, face images, and palmprints of campus users; The face feature recognition data includes facial feature indexes, facial contour indexes, skin texture gray values, and facial feature texture gray values; the fingerprint feature recognition data includes fingerprint shape indexes, ridge density, the number of bifurcation points, and the number of isolated points; the palmprint feature recognition data includes the number and length of main lines in the palmprint, the included angle between main lines, and the wrinkle density; Using Photoshop software, extract the fingerprint verification image gray value and the palmprint verification image gray value from the vertical images of the fingerprint instrument and the palmprint instrument respectively; Integrate the face feature recognition data, fingerprint feature recognition data, palmprint feature recognition data, fingerprint verification image gray value, palmprint verification image gray value, and the preprocessed payment data to generate a biometric verification data set, and divide the biometric verification data set into a training set and a test set.

5. The campus security payment system based on biometric recognition according to claim 4, wherein: The identity verification module constructs an identity verification model. The process of verifying the identity of campus users includes: Extract the face feature recognition data, fingerprint feature recognition data, and palmprint feature recognition data from the biometric verification data set; Combining the training set data with the multiple linear regression algorithm, using the face feature recognition data, fingerprint feature recognition data, and palmprint feature recognition data as inputs and the identity verification coefficient as the output, learn the linear relationship between the face feature recognition data, fingerprint feature recognition data, palmprint feature recognition data, and the identity verification coefficient, and train the identity verification model; Input the test set data into the identity verification model, adjust the intercept term and regression coefficient of the identity verification model, optimize the performance of the identity verification model, deploy the optimized identity verification model to the system, and combine the face feature recognition data, fingerprint feature recognition data, and palmprint feature recognition data to output the corresponding identity verification coefficient; Based on the identity verification coefficient, verify the identity of campus users to obtain the identity verification result. The identity verification result includes identity verification passed and identity verification failed.

6. The campus security payment system based on biometric recognition according to claim 5, wherein: The face recognition error analysis unit. The analysis process of the face recognition error rate includes: Normalize the face feature recognition data, and calculate the average values of the facial feature indexes, facial contour indexes, skin texture gray values, and facial feature texture gray values respectively; Assign weights to the calculated average values of the facial feature indexes, facial contour indexes, skin texture gray values, and facial feature texture gray values. Through the weights of the average values of the facial feature indexes and the weights of the average values of the facial contour indexes, obtain the face entry rate; through the weights of the average values of the skin texture gray values and the weights of the average values of the facial feature texture gray values, obtain the face texture gray value mutation rate; Set the thresholds of the facial feature indexes, facial contour indexes, and facial feature texture gray values respectively. Based on the set thresholds, calculate the abnormal face recognition rate; Assign weights to the obtained face entry rate, face texture gray value mutation rate, and abnormal face recognition rate respectively, adopt the weighted average method, analyze and obtain the face recognition error rate, and integrate the face recognition error rate into the biometric verification data set.

7. The campus security payment system based on biometric recognition according to claim 6, characterized in that: The fingerprint entry error analysis unit. The analysis process of the fingerprint entry error rate includes: Extract the grayscale values of fingerprint verification images in the biometric verification dataset. Using the training set data and combining with the linear regression algorithm, take the grayscale values of fingerprint verification images as the input and the fingerprint entry error rate as the output, learn the linear relationship between the grayscale values of fingerprint verification images and the fingerprint entry error rate, and train the fingerprint entry error analysis model; Input the test set data into the fingerprint entry error analysis model, adjust the parameters of the fingerprint entry error analysis model, optimize the fingerprint entry error analysis model, obtain the final fingerprint entry error analysis model, combine with the grayscale values of fingerprint verification images, output the corresponding fingerprint entry error rate, and integrate the fingerprint entry error rate into the biometric verification dataset.

8. The campus security payment system based on biometric recognition according to claim 7, characterized in that: For the palmprint entry error analysis unit, the analysis process of the palmprint entry error rate includes: Extract the grayscale values of palmprint verification images in the biometric verification dataset. Using the training set data and combining with the linear regression algorithm, take the grayscale values of palmprint verification images as the input and the palmprint entry error rate as the output, learn the linear relationship between the grayscale values of palmprint verification images and the palmprint entry error rate, and train the palmprint entry error analysis model; Input the test set data into the palmprint entry error analysis model, adjust the parameters of the palmprint entry error analysis model, optimize the palmprint entry error analysis model, obtain the final palmprint entry error analysis model, combine with the grayscale values of palmprint verification images, output the corresponding palmprint entry error rate, and integrate the palmprint entry error rate into the biometric verification dataset.

9. The campus security payment system based on biometric recognition according to claim 8, characterized in that: For the payment verification module, the process of constructing a payment verification model and then performing payment verification on campus users includes: Extract the payment data, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate in the biometric verification dataset; Using the training set data and combining with the multiple linear regression algorithm, take the payment data, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate as the input and the payment verification coefficient as the output, learn the linear relationship between the payment data, face recognition error rate, fingerprint entry error rate, palmprint entry error rate and the payment verification data, and train the payment verification model; Input the test set data into the payment verification model, adjust the intercept term and regression coefficient of the payment verification model, optimize the payment verification model, deploy the optimized payment verification model to the system, combine with the payment data, face recognition error rate, fingerprint entry error rate, and palmprint entry error rate, and output the corresponding payment verification coefficient; Based on the payment verification coefficient, perform payment verification on campus users to obtain the payment verification result, and the payment verification result includes payment verification passed and payment verification failed.

10. A campus security payment system based on biometric recognition according to claim 9, characterized in that: For the campus security payment module, the implementation process of the secure payment of campus users includes: S1. Set the identity verification as the first verification process and the payment verification as the second verification process; S2. Based on the identity verification result, when the identity verification of the campus user fails, do not perform the payment verification, switch the biometric recognition mode and perform the identity verification again; when the identity verification of the campus user passes, perform the payment verification; S3. Based on the payment verification result, when the payment verification of the campus user fails, the secure payment operation is not performed; when the payment verification of the campus user passes, the secure payment operation is performed.

Images