Identity authentication system based on biometric fingerprint identification technology

Abstract

The invention relates to the technical field of biological recognition, and discloses an identity authentication system based on a biological recognition fingerprint technology, and the system comprises a fingerprint data collection module which is used for collecting fingerprint two-dimensional image data and three-dimensional point cloud data through a multi-mode technology; the fingerprint feature extraction module is used for extracting feature vectors from the collected two-dimensional image data and three-dimensional point cloud data; the forgery detection module is used for analyzing dynamic pressing behaviors recorded in the collection process and detecting forgery fingerprints; the data privacy and security storage module is used for encrypting and storing the fingerprint feature template and executing feature matching in an encryption domain; and the identity authentication module is used for fusing the fingerprint features and other biological features. According to the method, the identity authentication precision and robustness are improved through the multi-modal dynamic feature fusion technology, the Paillier homomorphic encryption and dynamic behavior detection technologies are combined, data privacy protection is achieved, and the fake fingerprint recognition capability and the system security are remarkably enhanced.

Description

Technical Field

[0001] The present invention relates to the field of biometric identification technology, in particular to an identity authentication system based on biometric fingerprint technology. Background Art

[0002] With the rapid development of information technology and security needs, biometric-based identity authentication technology has become an important means to protect personal privacy and system security. Among them, fingerprint-based identity authentication systems are widely used in finance, government affairs, mobile device unlocking and other fields due to their uniqueness, ease of use and efficiency. Fingerprint authentication extracts the biometric information of the user's fingerprint and matches it with a pre-stored fingerprint template to complete identity authentication. It has the advantages of high accuracy and convenience.

[0003] Existing identity authentication systems based on biometric fingerprint technology usually use optical or capacitive sensors to collect fingerprint data and perform identity authentication through detail point matching or template comparison. Such systems excel in acquisition stability and matching accuracy, and can achieve efficient authentication, especially under clear fingerprint conditions.

[0004] However, existing technologies still have significant deficiencies when dealing with complex application scenarios. On the one hand, it is difficult to ensure the accuracy and robustness of authentication when single-modal features are collected under poor conditions or feature data is missing; on the other hand, traditional encryption methods still have the risk of being cracked during data storage and transmission, and the matching process does not provide sufficient protection for user privacy. In addition, the ability to identify forged fingerprints is limited, and it is unable to effectively defend against advanced forgery attacks. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides an identity authentication system based on biometric fingerprint technology, which solves the problems of insufficient multimodal feature fusion, imperfect privacy protection in the data storage and matching process, and weak forged fingerprint recognition capability existing in the prior art.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an identity authentication system based on biometric fingerprint technology, comprising: Fingerprint data collection module, used to collect fingerprint two-dimensional image data and three-dimensional point cloud data through multimodal technology; Fingerprint feature extraction module, used to extract feature vectors from collected two-dimensional image data and three-dimensional point cloud data; A forgery detection module, which is used to analyze the dynamic pressing behavior recorded during the acquisition process and detect forged fingerprints; A data privacy and security storage module, used to encrypt and store fingerprint feature templates and perform feature matching in an encrypted domain; The identity authentication module is used to integrate fingerprint features and other biometric features to perform identity authentication based on multimodal data.

[0007] Preferably, the fingerprint data collection module includes: Ultrasonic wave acquisition unit, used to calculate fingerprint three-dimensional point cloud data through ultrasonic wave reflection time; A multi-spectral acquisition unit, used to obtain optical reflectivity data of fingerprints at multiple skin layers through light of different wavelengths; The data fusion unit is used to combine the fingerprint three-dimensional point cloud data and the optical reflectivity data into a high-dimensional fingerprint template through a fusion algorithm.

[0008] Preferably, the ultrasonic acquisition unit calculates the three-dimensional point cloud data by the following formula:

[0009] in, is the height of the fingerprint surface, is the ultrasonic wave propagation speed, is the ultrasonic echo receiving time, is the ultrasonic emission time.

[0010] Preferably, the multi-spectral acquisition unit obtains optical reflectance data by the following formula:

[0011] in, is the wavelength of light, For the The reflectivity of the skin layer, is the attenuation coefficient of wavelength.

[0012] Preferably, the forgery detection module detects forged fingerprints by the following steps: Record dynamic pressing pressure during fingerprint collection p(t) and touch angle θ(t) ; Inputting the recorded dynamic pressing behavior into a forgery detection model for analysis, wherein the forgery detection model outputs a forgery probability based on time series analysis; When the forgery probability is greater than a preset threshold, it is determined to be a forged fingerprint and the authentication is rejected.

[0013] Preferably, the fingerprint feature extraction module generates a fingerprint feature template by the following steps: Extracting 2D image feature vectors through lightweight deep learning network and the 3D point cloud feature vector ; Generating high-dimensional feature templates based on weighted fusion formula :

[0014] in, and Dynamically adjust according to the reliability of 2D image and 3D point cloud features.

[0015] Preferably, the data privacy and security storage module includes: An encryption storage unit, used for encrypting and storing the fingerprint template by using Paillier homomorphic encryption technology; An encrypted matching unit, used for calculating the similarity of the encrypted feature template in the encrypted domain; Federated learning unit, used to collaboratively train fingerprint matching models across multiple devices in a distributed manner.

[0016] Preferably, the encryption matching unit calculates the similarity of the fingerprint features in the encryption domain by the following formula:

[0017] in, and They are the encrypted feature template to be matched and the stored template respectively.

[0018] Preferably, the identity authentication module includes: A multimodal fusion unit for fusing fingerprint features with iris features and facial features; The dynamic weight allocation unit is used to dynamically adjust the fusion weight according to the reliability of each modal feature.

[0019] Preferably, the multimodal fusion unit generates the authentication score by the following formula:

[0020] in, is a nonlinear activation function, is the dynamically assigned weight, defined as:

[0021] is the variance of the modal feature.

[0022] The present invention provides an identity authentication system based on biometric fingerprint technology. It has the following beneficial effects: 1. The present invention uses multimodal fusion technology to dynamically adjust the weights and fuse the multidimensional information of fingerprints, irises and facial features, thereby achieving high-precision identity authentication. The system can maintain the robustness and stability of authentication in complex environments, and effectively avoid the impact of incomplete or low-quality single-modal data.

[0023] 2. Through Paillier homomorphic encryption and federated learning technology, the present invention realizes efficient data privacy protection in the whole process of data storage, matching and model training, and can complete feature matching without decryption, thus ensuring the security of user data in the cloud and distributed environment.

[0024] 3. The present invention utilizes dynamic behavior detection and deep learning time series models to capture the physical characteristics of the fingerprint collection process in real time, achieving efficient forged fingerprint recognition. By combining multi-level feature collection with encrypted domain matching technology, the security and anti-attack capabilities of the authentication system are further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a system structure diagram of the present invention; Figure 2 It is a system diagram of the fingerprint data acquisition module of the present invention; Figure 3 It is a system diagram of the fingerprint feature extraction module of the present invention; Figure 4 It is a system diagram of the data privacy and security storage module of the present invention. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] Please refer to the attached Figure 1 -Attached Figure 4 The embodiment of the present invention provides an identity authentication system based on biometric fingerprint technology, including: Fingerprint data acquisition module, used to collect fingerprint two-dimensional image data and three-dimensional point cloud data through multimodal technology. The fingerprint data acquisition module improves the robustness and data quality of fingerprint acquisition, can work stably in harsh environments such as low light and humidity, and enhances the ability to resist attacks against forged fingerprints; Fingerprint feature extraction module, used to extract feature vectors from the collected 2D image data and 3D point cloud data. By combining the 2D image and 3D point cloud features, the feature extraction module improves the accuracy and robustness of fingerprint feature expression. Even when some fingerprint data is missing or of poor quality, it can still maintain a high feature extraction accuracy. The forgery detection module is used to analyze the dynamic pressing behavior recorded during the collection process and detect forged fingerprints. The forgery detection module enhances the system's anti-attack capability and can effectively prevent the system from being invaded by forged fingerprints, greatly improving the security of identity authentication; Data privacy and security storage module, which is used to encrypt and store fingerprint feature templates and perform feature matching in an encrypted domain. The data privacy and security storage module effectively protects the privacy of fingerprint feature templates. Even if the cloud server is attacked or the data is stolen, the attacker cannot restore the original fingerprint features, which significantly improves the security of the system. The identity authentication module is used to integrate fingerprint features and other biometric features and perform identity authentication based on multimodal data. By integrating multimodal features, the identity authentication module significantly improves the security and accuracy of the authentication system and can effectively deal with situations where a single biometric feature fails or is forged.

[0028] Please refer to the attached Figure 1 and attached Figure 2 In a preferred embodiment of the present invention, the fingerprint data acquisition module includes: Ultrasonic acquisition unit, used to calculate fingerprint three-dimensional point cloud data through ultrasonic reflection time. The ultrasonic acquisition unit can work stably in complex environments such as dry, wet, and contaminated. The generated three-dimensional point cloud data accurately reflects the height characteristics of the fingerprint, effectively improving the system's ability to process low-quality fingerprints; Multi-spectral acquisition unit, which is used to obtain optical reflectance data of fingerprints at multiple skin layers through light of different wavelengths. The multi-spectral acquisition unit can capture multi-level information of fingerprints, which not only enhances image quality, but also improves the system's ability to distinguish forged fingerprints, especially when detecting imitations such as silicone molds; The data fusion unit is used to combine the fingerprint 3D point cloud data and optical reflectivity data into a high-dimensional fingerprint template through a fusion algorithm. The ultrasonic point cloud data reflects the 3D geometric structure characteristics of the fingerprint, and the multispectral data reflects the epidermal and subcutaneous optical properties of the fingerprint. Through the deep fusion algorithm, the complementary information of the two data can be comprehensively utilized to generate a high-dimensional feature template, enhancing the expression ability and discrimination of fingerprint features.

[0029] Please refer to the attached Figure 1 and attached Figure 2 In a preferred embodiment of the present invention, the ultrasonic acquisition unit calculates the three-dimensional point cloud data by the following formula:

[0030] in, is the height of the fingerprint surface, is the ultrasonic wave propagation speed, is the ultrasonic echo receiving time, The ultrasonic emission time is the ultrasonic wave emission time. Through this principle, the ultrasonic wave acquisition unit can realize non-destructive and high-precision fingerprint acquisition. Especially under complex acquisition conditions (such as dry, wet or surface contamination), the ultrasonic wave's penetrating characteristics can effectively ignore surface interference, directly capture the real geometric features of the fingerprint, and ensure the reliability of the point cloud data.

[0031] Please refer to the attached Figure 1 and attached Figure 2 In a preferred embodiment of the present invention, the multi-spectral acquisition unit obtains optical reflectance data by the following formula:

[0032] in, is the wavelength of light, For the The reflectivity of the skin layer, is the attenuation coefficient of wavelength. Through the above formula, the multispectral acquisition unit can obtain detailed optical information of the epidermis and subcutaneous layer of the fingerprint in a single acquisition, and accurately capture the fingerprint texture, capillary structure and deep skin characteristics, providing high-quality multi-dimensional data for subsequent feature extraction. The multispectral acquisition unit has high robustness in complex acquisition environments and can provide rich multi-level biological information, which improves the accuracy of fingerprint acquisition and the ability to detect forged fingerprints.

[0033] Please refer to the attached Figure 1 In a preferred embodiment of the present invention, the forgery detection module detects forged fingerprints by the following steps: Record dynamic pressing pressure during fingerprint collection p(t) and touch angle θ(t) , use the embedded pressure sensor to detect the pressure change of the finger pressing area in real time. The pressure calculation formula is:

[0034] in, is the force change measured by the sensor during the acquisition process, is the contact area of ​​the pressed area; The tilt angle of the finger when pressing is detected by an embedded angle sensor or multiple arrayed sensors. The calculation formula for the touch angle is:

[0035] in, , , The maximum and minimum coordinates of the finger contact area are respectively. This step uses real-time pressure and angle changes to capture the dynamic behavior characteristics of the user during the pressing process and establishes physical property data of interaction with real fingers. These data are highly personalized and can effectively distinguish real fingers from forged fingerprints; The recorded dynamic pressing behavior is input into the forgery detection model for analysis. The forgery detection model outputs the forgery probability based on time series analysis. This step captures the time series characteristics in the pressing behavior through a deep learning model, effectively identifying the difference in dynamic behavior between forged fingerprints and real fingerprints. For example, forged fingerprints usually lack natural pressure and angle changes, thereby enhancing the accuracy of forgery detection; When the forgery probability is greater than the preset threshold, it is determined to be a forged fingerprint and the authentication is rejected. First, the forgery probability threshold is set. ,like =0.8, when the probability of forgery When the threshold is exceeded, it is judged as a forged fingerprint.

[0036] Through this judgment mechanism, the system can effectively reject authentication requests from forged fingerprints and prevent attackers from cracking the system through counterfeit fingerprints (such as silicone molds), thereby significantly improving the security of identity authentication.

[0037] Please refer to the attached Figure 1 and attached Figure 3 In a preferred embodiment of the present invention, the fingerprint feature extraction module generates a fingerprint feature template by the following steps: Extracting 2D image feature vectors through lightweight deep learning network and the 3D point cloud feature vector , extracting 2D and 3D features through a lightweight network not only improves the accuracy of feature extraction, but also greatly reduces the computational complexity, making it suitable for resource-constrained embedded devices and real-time processing scenarios. The extracted standardized features have high discrimination, providing high-quality input for subsequent fusion steps; Generating high-dimensional feature templates based on weighted fusion formula :

[0038] in, and The feature template generated by the dynamic adjustment based on the reliability of the 2D image and 3D point cloud features can adapt to the differences in data quality under different acquisition conditions. Even if the data quality of a certain mode of the 2D image or 3D point cloud is low, the weight adjustment mechanism can ensure the robustness and accuracy of the fusion template.

[0039] Please refer to the attached Figure 1 and attached Figure 4 In a preferred embodiment of the present invention, the data privacy and security storage module includes: An encryption storage unit, used for encrypting and storing the fingerprint template by using Paillier homomorphic encryption technology; An encrypted matching unit, used for calculating the similarity of the encrypted feature template in the encrypted domain; Federated learning unit, used to collaboratively train fingerprint matching models across multiple devices in a distributed manner.

[0040] The data privacy and security storage module achieves security protection of fingerprint templates through Paillier homomorphic encryption technology. Even if the storage database is obtained by an attacker, the attacker cannot decrypt the template, thereby ensuring the privacy of the user's fingerprint data. At the same time, the encrypted matching unit ensures data privacy protection during the matching process. Finally, the federated learning unit solves the privacy and security issues brought about by centralized storage of fingerprint data.

[0041] Please refer to the attached Figure 1 and attached Figure 4 In a preferred embodiment of the present invention, the encryption matching unit calculates the similarity of fingerprint features in the encryption domain by the following formula:

[0042] in, and They are the encrypted feature template to be matched and the stored template respectively. Through the above calculation formula, the encrypted matching unit completes the similarity calculation of the feature template in the encrypted domain, avoiding the data decryption process, thereby improving the data privacy protection capability in the matching process. At the same time, the encrypted matching unit realizes the feature matching process in the encrypted state through the homomorphic operation characteristics of Paillier encryption, ensuring the privacy and security of the data in the calculation process, and eliminating the need for frequent encryption and decryption operations, thereby optimizing the calculation efficiency.

[0043] Please refer to the attached Figure 1 In a preferred embodiment of the present invention, the identity authentication module includes: A multimodal fusion unit for fusing fingerprint features with iris features and facial features; Specifically: each modal feature is standardized to unify the dimensions and distribution of different modal features:

[0044] in, is the modal feature, and are the mean and standard deviation of the features respectively.

[0045] Use the weighted fusion formula to combine fingerprint features , iris features and facial features To perform the fusion:

[0046] in, , , The multimodal fusion unit combines three biometric features, fingerprint, iris and face, to effectively enhance the robustness and anti-attack ability of identity authentication.

[0047] A dynamic weight allocation unit is used to dynamically adjust the fusion weight according to the reliability of each modal feature; During operation, the system dynamically adjusts the weights based on the quality of the features collected in real time. For example, when the quality of the fingerprint features decreases (such as due to the collection environment), the system will reduce , and increase and The dynamic weight allocation unit automatically optimizes the fusion process by evaluating the reliability of modal features in real time, thereby improving the adaptability and stability of the authentication system and ensuring that a high authentication accuracy can be maintained when the quality of specific modal data is low.

[0048] Please refer to the attached Figure 1 In a preferred embodiment of the present invention, the multimodal fusion unit generates the authentication score by the following formula:

[0049] in, is a nonlinear activation function, is the dynamically assigned weight, defined as:

[0050] is the variance of the modal feature.

[0051] Specifically: Through the above formula, the multimodal fusion unit can efficiently and accurately integrate multimodal features to generate authentication scores, adapt to quality changes in different modal data, and provide consistent and reliable authentication results. Through the combination of dynamic weights and nonlinear activation functions, the system can achieve comprehensive utilization and precise mapping of multimodal features, ensuring the stability and discrimination of authentication scores.

[0052] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The identity authentication system based on biometric fingerprint technology is characterized by: include: Fingerprint data collection module, used to collect fingerprint two-dimensional image data and three-dimensional point cloud data through multimodal technology; Fingerprint feature extraction module, used to extract feature vectors from collected two-dimensional image data and three-dimensional point cloud data; A forgery detection module, which is used to analyze the dynamic pressing behavior recorded during the acquisition process and detect forged fingerprints; A data privacy and security storage module, used to encrypt and store fingerprint feature templates and perform feature matching in an encrypted domain; The identity authentication module is used to integrate fingerprint features and other biometric features to perform identity authentication based on multimodal data.

2. The identity authentication system based on biometric fingerprint technology according to claim 1 is characterized in that: The fingerprint data acquisition module comprises: Ultrasonic wave acquisition unit, used to calculate fingerprint three-dimensional point cloud data through ultrasonic wave reflection time; A multi-spectral acquisition unit, used to obtain optical reflectivity data of fingerprints at multiple skin layers through light of different wavelengths; The data fusion unit is used to combine the fingerprint three-dimensional point cloud data and the optical reflectivity data into a high-dimensional fingerprint template through a fusion algorithm.

3. The identity authentication system based on biometric fingerprint technology according to claim 2 is characterized in that: The ultrasonic acquisition unit calculates the three-dimensional point cloud data by the following formula: in, is the height of the fingerprint surface, is the ultrasonic wave propagation speed, is the ultrasonic echo receiving time, is the ultrasonic emission time.

4. The identity authentication system based on biometric fingerprint technology according to claim 2 is characterized in that: The multi-spectral acquisition unit obtains optical reflectance data through the following formula: in, is the wavelength of light, For the The reflectivity of the skin layer, is the attenuation coefficient of wavelength.

5. The identity authentication system based on biometric fingerprint technology according to claim 1, characterized in that: The forgery detection module detects forged fingerprints by the following steps: Record dynamic pressing pressure during fingerprint collection p(t) and touch angle θ(t) ; Inputting the recorded dynamic pressing behavior into a forgery detection model for analysis, wherein the forgery detection model outputs a forgery probability based on time series analysis; When the forgery probability is greater than a preset threshold, it is determined to be a forged fingerprint and the authentication is rejected.

6. The identity authentication system based on biometric fingerprint technology according to claim 1, characterized in that: The fingerprint feature extraction module generates a fingerprint feature template by the following steps: Extracting 2D image feature vectors through lightweight deep learning network and the 3D point cloud feature vector ; Generating high-dimensional feature templates based on weighted fusion formula : in, and Dynamically adjust according to the reliability of 2D image and 3D point cloud features.

7. The identity authentication system based on biometric fingerprint technology according to claim 1, characterized in that: The data privacy and security storage module includes: An encryption storage unit, used for encrypting and storing the fingerprint template by using Paillier homomorphic encryption technology; An encrypted matching unit, used for calculating the similarity of the encrypted feature template in the encrypted domain; Federated learning unit, used to collaboratively train fingerprint matching models across multiple devices in a distributed manner.

8. The identity authentication system based on biometric fingerprint technology according to claim 7, characterized in that: The encryption matching unit calculates the similarity of fingerprint features in the encryption domain using the following formula: in, and They are the encrypted feature template to be matched and the stored template respectively.

9. The identity authentication system based on biometric fingerprint technology according to claim 1, characterized in that: The identity authentication module comprises: A multimodal fusion unit for fusing fingerprint features with iris features and facial features; The dynamic weight allocation unit is used to dynamically adjust the fusion weight according to the reliability of each modal feature.

10. The identity authentication system based on biometric fingerprint technology according to claim 1, characterized in that: The multimodal fusion unit generates the certification score by the following formula: in, is a nonlinear activation function, is the dynamically assigned weight, defined as: is the variance of the modal feature.

Images