DE202023103427U1 - A system for securely updating biometric feature vectors using the Age Adaptation Feature Vector on blockchain - Google Patents

Abstract

A system (100) for securely updating the biometric feature vector using the age matching feature vector on blockchain, the system (100) comprising:
a multimodal biometric authentication module (102) configured to capture multiple biometrics, such as a combination of an invariant biometric and a variant biometric of a user;
a change detection module (104) coupled to the multimodal biometric authentication module (102) for analyzing the plurality of captured biometric characteristics over time and detecting changes in the characteristics based on age, health or environmental factors;
a biometric verification module (106) coupled to the change detection module (104) and configured to use the biometrics to verify a user's identity based on the detected changes in the captured plurality of biometrics; And
a blockchain network (108) connected to the biometric verification module (106) and configured to store user registration details and timestamps.

Description

FIELD OF THE INVENTION

The present invention relates to the field of blockchain systems. In particular, the present subject matter relates to a system for securely updating biometric feature vectors using the Age Adaptation Feature Vector on blockchain.

BACKGROUND OF THE INVENTION

With the rapid development of the Internet and mobile devices, authentication systems are widely used when accessing Internet services and mobile devices to protect user devices, content and accounts.

Biometrics is the measurement of a physical or biological characteristic or attribute of an individual to confirm or determine that individual's identity.

Biometric systems are typically used in applications where secure access is required. Secure access can relate to physical locations (e.g., parts of a building), electronic devices and systems (e.g., ATMs, cell phones, personal computers), and software applications.

Secure access can be accomplished by a user entering data such as login codes or passwords. However, these can be compromised relatively easily if a third party gains knowledge of the data to be entered. Secure access can also be guaranteed by combinations of an ID card and a personal identification number (PIN), such as is often required at ATMs (cash dispensers). However, this can also be compromised if the card details and PIN are obtained from a third party.

Biometric systems have the advantage of measuring a characteristic of an individual to determine whether that individual should be allowed access to a secure area or device. This makes access by third parties considerably more difficult, since no data is entered. However, biometric systems have the disadvantage that it is relatively difficult to obtain accurate readings from an individual, particularly when the individual is a member of the public who has not been trained in the use of the biometric system and when the system being used in an unattended environment.

Biometric recognition is advantageous over token-based authentication because the focus is on "who the person is" rather than "what the person owns". Any physiological or behavioral feature can be used to create a classifier. Each biometric feature is further categorized into variant and invariant biometric features. Variant biometrics are those that change over time due to aging or other physiological or environmental conditions, while invariant biometrics are not affected by any of these situations. Unimodal biometric systems suffer from noisy sensor data, non-universality, lack of individuality and the unavailability of invariant representations. However, multimodal biometric authentication systems have proven to be more reliable, robust, and accurate than unimodal authentication systems. It makes sense to set up a multimodal biometric authentication system with at least one invariant biometric feature. Features such as iris and fingerprint with high persistence are considered invariant biometric features.

US7493496B2 discloses a biometric system for assisting a user in providing a biometric reading. The system includes: a biometric acquisition unit for acquiring biometric data of a user; and feedback means for providing the user with an indication of the extent to which a measurement has been captured. The feedback means can be implemented by a display, a speaker, or both. The system can also comprise anonymization means for providing the feedback means with an anonymized version of the data measured by the biometric detection unit.

US7298873B2 discloses a multimodal biometric identification or authentication system that includes a plurality of biometric clients. Each of the biometric clients can use devices to Capturing biometric images of several types. The system includes a router that communicates with the biometric clients. The router receives biometric images from the biometric clients and sends biometric scores or results back to them. The system includes multiple biometric matching engines that communicate with the router. Each biometric matching engine includes multiple biometric processors. Each biometric processor is capable of processing biometric data of a specific type. The biometric matching engines transmit and receive biometric data to and from the router.

Most of the techniques introduced and models built for age-invariant face recognition are based on datasets containing face samples with advancing age.

The first technique is to build a robust biometric authentication system that focuses on enrichment of feature vectors from different biometric traits. Boussaad, L., Benmohammed, MBenzid, and Rhave presented their way of solving the same problem in faces that age over time. They used the Active Appearance Model (AAM) to generate an AAM-based facial representation. A two-dimensional discrete cosine transform (2D-DCT) was then applied to generate an image descriptor, and then Kernel-Fisher analysis (KFA) was applied for dimension reduction. Classification by a nearest neighbor K classifier based on Euclidean distance.

According to one researcher, Anupa Sinha, Dr. Snehlata Barde four phases: face detection, pre-processing, highlight extraction and face detection. They used Viola Jones calculations for face detection and in the pre-processing phase the Gabor channel is used to eliminate noise from the image. They used PCA and mSVM to extract highlights from images and detect faces, respectively. mSVM is a classifier for which PCA is the contribution to the type of highlights and perceived faces.

According to another researcher, Hachim El Khiyari, Harry Wechsler proposed a combined method of generative and discriminative methods through the medium of transfer learning. To address the same problem, existing methods of facial aging can be divided into two main groups: generative and discriminative. Generative methods typically rely on statistical models to predict the appearance of faces at different target ages. On the other hand, discriminative methods avoid building a model for facial aging as generative methods would. They attempt to match images directly for authentication, without the intermediate step of creating synthetic faces. Age invariance can be implemented at either the feature extraction, training and/or recognition level. At the feature extraction level, the goal is to derive image descriptors that are robust to intrapersonal aging variations.

According to another researcher, Lanitis et al. developed a generative statistical model that allows the simulation or elimination of aging effects in facial images. Ling et al. used Gradient Orientation Pyramids (GOP) by extracting the directions of the gradient vectors at multiple scales while discarding the magnitude components. At the training and testing level, one strives for robust generalization despite aging through learning. In Biswas et al. Aging was addressed at the detection level by analyzing and measuring face drift due to age progression. When two images show the same subject, the variance is coherent, while for images of different subjects, the variance is extreme or incoherent.

Many existing methods have attempted to build models using the age-invariant face recognition database, but did not take into account the subject-related confidence level with which the recognition is accepted and rejected. Some methods have placed an emphasis on the key descriptors of the face that could help avoid false rejections when the system is used over a long period of time. But in sensitive applications such as banking, military, etc., it is always advisable to collect biometric feature samples from subjects to ensure the non-repudiation, security and vitality of the subject.

However, none of the above prior art discloses a secure biometric system that resists attack and securely stores the biometric information. Each physiological and/or behavioral feature is used to build unimodal or multimodal biometric recognition systems. In order to meet the authentication criteria such as universality, uniqueness, to meet permanence and acceptance; A multimodal biometric system is preferred to a unimodal biometric system. The multimodal biometric authentication system is built by combining feature vectors from more than one biometric feature. The process of selecting biometric features to generate combined feature vectors must be intuitive. Physiological biometric features such as the face can change over time due to wrinkles, obesity, etc. Likewise, behavioral characteristics such as voice and signature also change due to aging. These trait-based models can lead to an increased rate of false rejection when used over a long period of time, as shown in the figure. This problem is evident in almost all biometric authentication systems and is critical for those who prefer cancelable biometrics.

Therefore, in order to overcome the limitations of the above prior art, there is a need to develop a secure multi-modal biometric system for securely updating biometric feature vectors using the age matching feature vector on blockchain.

The technical advances disclosed by the present invention overcome the limitations and disadvantages of existing and conventional systems and methods.

SUMMARY OF THE INVENTION

The present invention relates generally to a system for securely updating biometric feature vectors using age matching feature vectors on the blockchain.

An object of the present invention is to provide a multimodal biometric system.

Another object of the present invention is to use a blockchain network for securely updating the biometric feature vector using the age matching feature vector.

Another aim of the present invention is to detect changes in various biometric characteristics, monitor the events and execute smart contracts.

Another object of the present invention is to implement a smart contract to trigger re-enrollment of different biometrics.

Another object of the present invention is to use blockchain to ensure security, non-repudiation and availability of updated feature vectors by storing the logs.

• ein multimodales biometrisches Authentifizierungsmodul, das so konfiguriert ist, dass es mehrere biometrische Merkmale erfasst, beispielsweise eine Kombination aus einem invarianten biometrischen Merkmal und einem abweichenden biometrischen Merkmal eines Benutzers;
• ein Änderungserkennungsmodul, das mit dem multimodalen biometrischen Authentifizierungsmodul verbunden ist, um die Vielzahl der erfassten biometrischen Merkmale über einen Zeitraum zu analysieren und Änderungen in den Merkmalen basierend auf Alterung, Gesundheit oder Umweltfaktoren zu erkennen;
• ein biometrisches Verifizierungsmodul, das mit dem Änderungserkennungsmodul verbunden ist, das für die biometrischen Merkmale konfiguriert ist, um die Identität eines Benutzers basierend auf den erkannten Änderungen in der erfassten Vielzahl biometrischer Merkmale zu überprüfen; Und
• ein Blockchain-Netzwerk, das mit dem biometrischen Verifizierungsmodul verbunden ist und so konfiguriert ist, dass es Details und Zeitstempel der Benutzerregistrierung im biometrischen Authentifizierungsmodul speichert.

In one embodiment, a system for securely updating biometric feature vectors using the age matching feature vector on blockchain, the system comprising:

• a multimodal biometric authentication module configured to capture multiple biometrics, such as a combination of an invariant biometric and a variant biometric of a user;
• a change detection module connected to the multimodal biometric authentication module to analyze the plurality of captured biometric features over time and detect changes in the features based on aging, health or environmental factors;
• a biometric verification module coupled to the change detection module configured for the biometrics to verify a user's identity based on the detected changes in the captured plurality of biometrics; And
• a blockchain network connected to the biometric verification module and configured to store user registration details and timestamps in the biometric authentication module.

In one embodiment, a notification module is coupled to the change detection module to notify the administrator of any updates made to a user's biometric registration based on the detected changes and a confidence score.

• Erfassen Sie das Paar biometrischer Merkmale mit mindestens hoher Beständigkeit, wie z. B. einer Iris, oder geringer Beständigkeit, wie z. B. der Unterschrift.
• Extrahieren von mindestens einem Merkmalsvektor aus dem erfassten Paar biometrischer Merkmale durch ein Merkmalsextraktionsmodul; Und
• Klassifizieren Sie mithilfe eines multimodalen Klassifizierungsmoduls die extrahierten Merkmalsvektoren, um die Identität eines Benutzers über einen bestimmten Zeitraum und ein Konfidenzniveau basierend auf einem Faltungs-Neuronalen Netzwerk und einem Langzeit-Kurzzeitgedächtnis (LSTM) zu bestimmen.

In one embodiment, the multimodal biometric authentication module is configured to:

• Capture the pair of biometrics with at least high consistency, such as B. an iris, or low permanence, such. B. the signature.
• extracting at least one feature vector from the detected pair of biometric features by a feature extraction module; And
• Using a multimodal classification engine, classify the extracted feature vectors to determine a user's identity over a specified time period and confidence level based on a convolutional neural network and long-term short-term memory (LSTM).

In one embodiment, the extracted features are combined together to form a pair of unimodal modules for each pair of biometric features.

In one embodiment, an initialization module is connected to the feature extraction module to define a maximum initial low confidence threshold as an average confidence value resulting from validation of the input and to set a threshold time for updating the feature vector of the variant biometric feature to a period more than 6 months from a current date.

In one embodiment, a comparison module is coupled to the initialization module and the multi-modal biometric authentication module to compare the obtained confidence to the initialized confidence value and the obtained time to the threshold time.

In one embodiment, an update module is coupled to the comparison module and the blockchain network to trigger a smart contract to update the feature vector, wherein the update module is triggered when either the confidence is less than the confidence threshold or the threshold time is less than either greater than or equal to the initialized threshold time.

In order to further clarify the advantages and features of the present invention, a more detailed description of the invention will be given with reference to specific embodiments thereof illustrated in the accompanying drawings. It should be understood that these drawings represent only typical embodiments of the invention and are therefore not to be considered as limiting the scope thereof. The invention is described and explained in greater detail with reference to the accompanying drawings.

character list

• 1 zeigt ein Blockdiagramm eines Systems zur sicheren Aktualisierung biometrischer Merkmalsvektoren mithilfe des Age Adaptation Feature Vector auf Blockchain
• 2 zeigt ein Flussdiagramm des vorgeschlagenen multimodalen biometrischen Authentifizierungssystems.

These and other features, aspects and advantages of the present invention will be better understood when the following detailed description is read with reference to the accompanying drawings, in which like reference characters represent like parts throughout the drawings, wherein:

• 1 shows a block diagram of a system for securely updating biometric feature vectors using the Age Adaptation Feature Vector on blockchain
• 2 shows a flowchart of the proposed multimodal biometric authentication system.

In addition, skilled craftsmen will recognize that elements in the drawings are presented for simplicity and may not necessarily have been drawn to scale. For example, the flowcharts illustrate the method with key steps that help improve understanding of aspects of the present disclosure. Furthermore, in view of the construction of the device, one or more components of the device may have been represented in the drawings by conventional symbols, and the drawings may only show the specific details necessary for an understanding of the embodiments of the present disclosure tion are relevant so as not to obscure the drawings with details that are readily apparent to those of ordinary skill in the art, having the benefit of the description herein.

DETAILED DESCRIPTION:

For the purposes of promoting an understanding of the principles of the invention, reference will now be made to the embodiment illustrated in the drawings and specific language will be used to describe the same. It should be understood, however, that no limitation on the scope of the invention is intended thereby, as variations and further modifications to the illustrated system and further applications of the principles of the invention illustrated therein are contemplated as would normally occur to one skilled in the art Technique to which the invention relates.

Those skilled in the art will understand that the foregoing general description and the following detailed description are exemplary and explanatory of the invention and are not intended to be limiting.

Reference throughout this specification to "an aspect," "another aspect," or similar language means that a particular feature, structure, or feature described in connection with the embodiment is in at least one embodiment of the present invention is included. As such, the phrases "in one embodiment," "in another embodiment," and similar phrases throughout this specification may, but do not necessarily, refer to the same embodiment.

The terms "comprises," "comprising," or any other variation thereof is intended to cover non-exclusive inclusion such that a process or method that includes a list of steps includes not only those steps, but may include other steps not expressly listed or inherent in that process or method. Likewise, any device or subsystem or element or structure or component preceded by “includes...a” does not exclude, without further limitation, the existence of other devices or other subsystems or other elements or other structure other components or additional devices or additional subsystems or additional elements or additional structures or additional components.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The system, methods, and examples provided herein are for purposes of illustration only and are not intended to be limiting.

Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

• ein multimodales biometrisches Authentifizierungsmodul (102), das so konfiguriert ist, dass es mehrere biometrische Merkmale erfasst, beispielsweise eine Kombination aus einem invarianten biometrischen Merkmal und einem abweichenden biometrischen Merkmal eines Benutzers;
• ein Änderungserkennungsmodul (104), das mit dem multimodalen biometrischen Authentifizierungsmodul (102) verbunden ist, um die Vielzahl der erfassten biometrischen Merkmale im Zeitverlauf zu analysieren und Änderungen in den Merkmalen basierend auf Alter, Gesundheit oder Umweltfaktoren zu erkennen;
• ein biometrisches Verifizierungsmodul (106), das mit dem Änderungserkennungsmodul (104) verbunden ist und so konfiguriert ist, dass es die biometrischen Merkmale verwendet, um die Identität eines Benutzers auf der Grundlage der erkannten Änderungen in der erfassten Vielzahl biometrischer Merkmale zu überprüfen; Und
• ein Blockchain-Netzwerk (108), das mit dem biometrischen Verifizierungsmodul (106) verbunden ist und zum Speichern von Details und Zeitstempeln der Benutzerregistrierung konfiguriert ist.

1 shows a block diagram of a system (100) for securely updating the biometric feature vector using the age adaptation feature vector on blockchain, the system (100) comprising:

• a multimodal biometric authentication module (102) configured to capture multiple biometrics, such as a combination of an invariant biometric and a variant biometric of a user;
• a change detection module (104) coupled to the multimodal biometric authentication module (102) for analyzing the plurality of captured biometric features over time and detecting changes in the features based on age, health or environmental factors;
• a biometric verification module (106) coupled to the change detection module (104) and configured to use the biometrics to verify a user's identity based on the detected changes in the captured plurality of biometrics; And
• a blockchain network (108) connected to the biometric verification module (106) and configured to store user registration details and timestamps.

In one embodiment, a notification module (110) is coupled to the change detection module (104) to notify the administrator of any updates made to the biometric Registration of a user can be made based on the detected changes and a confidence value.

• Erfassen Sie das Paar biometrischer Merkmale mit mindestens hoher Beständigkeit, wie z. B. einer Iris, oder geringer Beständigkeit, wie z. B. der Unterschrift.
• Extrahieren von mindestens einem Merkmalsvektor aus dem erfassten Paar biometrischer Merkmale durch ein Merkmalsextraktionsmodul (102a); Und
• Klassifizieren Sie die extrahierten Merkmalsvektoren durch ein multimodales Klassifizierungsmodul (102b), um die Identität eines Benutzers über einen Schwellenwertzeitraum und ein Konfidenzniveau basierend auf einem Faltungs-Neuronalen Netzwerk und einem Langzeit-Kurzzeitgedächtnis (LSTM) zu bestimmen.

Figure 2 shows a flowchart of the proposed multimodal biometric authentication system. The multimodal biometric authentication module (102) is configured for:

• Capture the pair of biometrics with at least high consistency, such as B. an iris, or low permanence, such. B. the signature.
• extracting at least one feature vector from the detected pair of biometric features by a feature extraction module (102a); And
• Classify the extracted feature vectors by a multimodal classification module (102b) to determine a user's identity over a threshold period and confidence level based on a convolutional neural network and long-term short-term memory (LSTM).

In one embodiment, the extracted features are combined together to form a pair of unimodal modules for each pair of biometric features.

In one embodiment, an initialization module (112) is coupled to the feature extraction module (102a) to define a maximum initial low confidence threshold as an average confidence value resulting from validation of the input and to set a threshold time for the feature vector to be updated if it is different biometric feature for a period of more than 6 months from a current point in time.

In one embodiment, a comparison module (114) is coupled to the initialization module (112) and the multimodal biometric authentication module (102) to compare the obtained confidence to the initialized confidence value and the obtained time to the threshold time.

In one embodiment, an update module (116) is connected to the comparison module (114) and the blockchain network (108) to trigger a smart contract to update the feature vector, the update module (116) being triggered when the confidence is less than the threshold and threshold time is either greater than or equal to the initialized threshold time.

1. A. Invariantes biometrisches Merkmal: Ein multimodaler Klassifikator muss einen Merkmalsvektor von mindestens einem biometrischen Merkmal enthalten, der über die Zeit konstant bleibt. Dieser Merkmalsvektor wird verwendet, um die Aktualisierung von Varianten-Merkmalsvektoren zu fördern. Invariante biometrische Merkmale sind Fingerabdruck, Handflächenabdruck, Iris und Netzhautscan.
2. B. Variante des biometrischen Merkmals: Ein biometrisches Merkmal, dessen Merkmalsvektor über die Zeit nicht konstant bleibt, wird als Variante des biometrischen Merkmals bezeichnet. Invariante Merkmalsvektoren werden verwendet, um die Aktualisierung von Varianten-Merkmalsvektoren zu fördern.
3. C. Zulässiger Varianzgrad im Konfidenzniveau: Zulässiger Varianzgrad der biometrischen Merkmale, für die keine Aktualisierung erforderlich ist oder die Leistungsmetriken innerhalb der erforderlichen Grenzen liegen. Bei biometrischen Merkmalen wie Unterschrift und Stimme wird diese Varianz anhand des euklidischen Distanzmodells berechnet, und für Gesichtsmerkmale wird die Gesichtsregressionstechnik verwendet.
4. D. Multimodales biometrisches Modell: Die Fusion wird auf Entscheidungsebene durchgeführt, um einen Klassifikator zur Durchführung der biometrischen Authentifizierung zu erstellen. Dieses vorgeschlagene System besteht aus drei Kanälen, wobei zwei Kanäle einem ausgewählten invarianten Merkmal und einem Variantenmerkmal entsprechen und der dritte dem kombinierten Merkmalsvektor dieser beiden entspricht.
5. E. Zeitpunkt zum Auslösen der Aktualisierung des Feature-Vektors: Zu jedem Zeitpunkt wird das Ereignis zum Aktualisieren des Feature-Vektors ausgelöst, wenn eine der folgenden Bedingungen erfüllt ist.
   1. i. Zeit zum Aktualisieren des Feature-Vektors: Wenn die bereits festgelegte Zeit abläuft, wird ein Ereignis zur Aktualisierung des Feature-Vektors ausgelöst. Derzeit ist sie auf 6 Monate festgelegt. Es kann auf jeden logischen Zeitpunkt geändert werden, indem das System über einen längeren Zeitraum als 6 Monate analysiert wird. Es kann vorkommen, dass Merkmalsvektoren aktualisiert werden, auch wenn der Schwellenwert für die maximale niedrige Konfidenzbewertung nicht erreicht ist.
   2. ii. Schwellenwert für niedrige Konfidenzniveauzahl: Die Häufigkeit, mit der das Subjekt mit einem niedrigeren Konfidenzwert als dem festgelegten Schwellenwert authentifiziert wird oder wie oft das Subjekt nicht authentifiziert wird, wenn ein invariantes Merkmal das Subjekt authentifiziert hat, den festgelegten Schwellenwert überschreitet, der den maximalen niedrigen Konfidenzwert darstellt, dann kann es zu einem Aktualisierungsereignis kommen des Merkmalsvektors, obwohl noch keine Zeit verstrichen ist.

The system has a biometric feature that does not change over time, such as B. Fingerprint, handprint, iris, DNA, etc. In order to recognize the change of biometric feature (variant biometric feature) such as face signature, a multi-modal biometric authentication system is introduced. This system detects changes in the variant's biometrics due to various factors such as aging, environmental changes, health problems, etc., monitors these events over time and, according to set parameters, executes a smart contract to trigger the re-registration of the variant's biometrics. Events are logged on a private blockchain for secure updating and future auditing purposes.

1. A. Invariant Biometric Feature: A multimodal classifier must contain a feature vector of at least one biometric feature that remains constant over time. This feature vector is used to promote updating of variant feature vectors. Invariant biometric features are fingerprint, palmprint, iris and retinal scan.
2. B. Biometric Feature Variant: A biometric feature whose feature vector does not remain constant over time is referred to as a biometric feature variant. Invariant feature vectors are used to promote updating of variant feature vectors.
3. C. Permissible degree of variance in confidence level: Permissible degree of variance of the biometrics for which no update is required or the performance metrics are within the required limits. For biometric features such as signature and voice, this variance is calculated using the Euclidean distance model, and for facial features, the facial regression technique is used.
4. D. Multimodal Biometric Model: Fusion is performed at decision level to create a classifier to perform biometric authentication. This proposed system consists of three channels, two channels corresponding to a selected invariant feature and a variant feature, and the third corresponding to the combined feature vector of these two.
5. E. Time to trigger feature vector update: At any time, the feature vector update event will be triggered if any of the following conditions are met.
   1. i. Time to update feature vector: When the time already set expires, a feature vector update event is dispatched. It is currently set at 6 months. It can be changed to any logical point in time by analyzing the system over a period longer than 6 months. It can happen that feature vectors are updated even if the threshold for the maximum low confidence score is not reached.
   2. ii. Low Confidence Level Number Threshold: The number of times the subject is authenticated with a confidence score lower than the specified threshold, or the number of times the subject is not authenticated when an invariant characteristic has authenticated the subject exceeds the specified threshold, which is the maximum low confidence score represents, then an update event of the feature vector can occur even though time has not yet elapsed.

To trigger feature vector updates, biometric authentication systems are integrated into the blockchain. Depending on the outcome of these three channels, an appropriate smart contract is invoked. Table 1: Details of smart contracts to be invoked depending on the output of deep learning models M1O _{_} _ M2O _{_} _ M12O _{_} _ Smart Contract Authenticate Reject/authenticate with low confidence Authenticate Record the date and time and increase the number of false rejections of the subject. Authenticate Authenticate Authenticate In order to accomplish certain tasks, authenticated parties want to be involved. eg vote

A pair of biometric features, ie a high permanence iris and a low permanence signature, is selected to construct the unimodal deep learning models M _{1 and M 2 from the iris and signature feature vectors fv 1 and fv 2} , respectively

These feature vectors are combined to create a multimodal classifier M ₁₂

Set a threshold time_to_update_feature_vector to 6 months later datetime from the current one.

Set the maximum low_confidence_level threshold to the average confidence_score resulting from validating the inputs into the biometric authentication system.

• Das anfängliche Vertrauen wird wie folgt definiert:
  • Angenommen, wir haben n Proben aus der Klasse C_i. Der Merkmalsvektor von n Stichproben wird als { fv₀, fv₁,fv₂,....fv_n-1} dargestellt.

To define the confidence threshold for each subject:

• Initial trust is defined as follows:
  • Suppose we have n samples from class C _i . The feature vector of n samples is represented as { fv ₀ , fv ₁ ,fv ₂ ,....fv _n-1 }.

Calculate the Euclidean distance between fv ₀ , with fv ₁ to fv _n-1 , as {dv _i1 , dv _i2 ,.... dv _in-1 }

Calculate the mean: $$\underset{\_}{\mathrm{i.e}{v}_i}=\frac{\mathrm{i.e}{v}_{i1}+\mathrm{i.e}{v}_{i2}+\cdots .+\mathrm{i.e}{v}_{in-1}}{n}$$

Calculate the sample variance for the new sample from the ith class: $$s_i=\sqrt{\frac{{{\displaystyle \sum \left(\mathrm{i.e}{v}_{new}-\underset{\_}{\mathrm{i.e}{v}_i}\right)}}^2}{n-1}}$$

Where dv _new is the average feature vector distance of the feature vector of a new sample from the feature vector of the other samples from the i-th class.

dann ist das (1-a)%-Konfidenzintervall für die Varianz σ²: $$\left(\frac{\left(n-1\right).s_i^2}{b}\right)\le {\sigma }^2\le \left(\frac{\left(n-1\right),s_i^2}{a}\right)$$

If x ₁ , x ₂ ......x _n are normally distributed and $a={\chi }_{1-\frac{a}{\mathrm{2,}n-1}}^2\text{and}b={\chi }_{\frac{a}{\mathrm{2,}n-1}}^2,$

then the (1-a)% confidence interval for the variance σ ² is: $$\left(\frac{\left(n-1\right).s_i^2}{b}\right)\le {\sigma }^2\le \left(\frac{\left(n-1\right),s_i^2}{a}\right)$$

This is an initial confidence interval for classification, recording the minimum confidence for each test and using the last five values to define the maximum low_confidence level for the next test.

Input the input test sample to the biometric authentication system consisting of three models and record its output as {M ₁ O, M ₂ O, M ₁₂ O}.

• ii. SmartContracti - Sobald das Subjekt authentifiziert wird , wird es zur eigentlichen Anwendung weitergeleitet, die auf der Blockchain ausgeführt wird. ZB um an einer Wahl teilzunehmen.
• iii. SmartContract2 - Erfassen Sie falsche Ablehnungen für jedes Subjekt , indem Sie die Anzahl von fr erhöhen, indem Sie das Konfidenzniveau überprüfen. Wenn das Subjekt mit einem Konfidenzniveau unterhalb des Schwellenwerts authentifiziert wird, wird dies intern als falsche Ablehnung gemeldet und fr um eins erhöht.
• iv. SmartContract3 - Vergleicht fr mit nfr. Wenn fr >= nfr, wird das Subjekt benachrichtigt, biometrische Merkmalsproben erneut zu registrieren.

SmartContract1 - Redirect to functionality implemented in the blockchain to complete the task/activity.

• ii. SmartContracti - Once the subject is authenticated, it is passed to the actual application running on the blockchain. For example, to take part in an election.
• iii. SmartContract2 - Capture false rejections for each subject by increasing the number of fr by checking the confidence level. If the subject is authenticated with a confidence level below the threshold, this is internally reported as a false rejection and fr increased by one.
• IV. SmartContract3 - Compares fr with nfr. If fr >= nfr, the subject is notified to re-enroll biometric feature samples.

The drawings and the above description give examples of embodiments. Those skilled in the art will recognize that one or more of the elements described may well be combined into a single functional element. Alternatively, certain elements can be broken down into multiple functional elements. Elements of one embodiment may be added to another embodiment. For example, the orders of the processes described herein may be changed and are not limited to the manner described herein. Additionally, the actions of a flowchart need not be implemented in the order shown; Not all actions have to be carried out. Actions that are not dependent on other actions can also be performed in parallel with the other actions. The scope of the embodiments is in no way limited by these specific examples. Numerous variations, whether explicitly stated in the specification or not, such as B. Differences in structure, dimensions and material use are possible. The scope of the embodiments is at least as broad as indicated by the following claims.

Advantages, other benefits, and solutions to problems have been described above in terms of specific embodiments. However, the advantages, benefits, problem solutions, and any component that may cause any benefit, advantage, or solution to occur or become more pronounced, should not be construed as a critical, required, or essential function or component of any or all of the claims.

Reference List

100

A system (100) for securely updating the biometric feature vector using the age matching feature vector on the blockchain.

102

Multimodal biometric authentication module

102a

feature extraction module

102b

Multimodal classification module

104

change detection module

106

biometric verification module

108

blockchain network

110

notification module

112

initialization module

114

comparison module

116

update module

202

Registration of invariant biometric features (iris)

204

Biometric authentication module

206

Registration of various biometric features (signature)

208

Trigger the Update_feature_vector smart contract

210

If datetime >= time to_update _feature_vector

212

Record the confidence and compare it to low_confidence_threshold , record date and time, from M2O

214

If confidence < low_confidence_threshold,

216

Initial define low_confidence_threshold and time_to_update feature_vector

218

Feature vector extraction, fv2

220

Feature vector extraction, fv1

222

Convolutional Neural Network

224

Combined fv1 and fv2 based model

226

perform classification

228

Record Confidence CF

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• US 7493496 B2 [0008]
• US 7298873 B2 [0009]

Claims (7)

A system (100) for securely updating the biometric feature vector using the age matching feature vector on blockchain, the system (100) comprising: a multimodal biometric authentication module (102) configured to capture multiple biometrics, such as a combination of an invariant biometric and a variant biometric of a user; a change detection module (104) coupled to the multimodal biometric authentication module (102) for analyzing the plurality of captured biometric features over time and detecting changes in the features based on age, health or environmental factors; a biometric verification module (106) coupled to the change detection module (104) and configured to use the biometrics to verify a user's identity based on the detected changes in the captured plurality of biometrics; And a blockchain network (108) connected to the biometric verification module (106) and configured to store user registration details and timestamps. system after claim 1 wherein a notification module (110) is coupled to the change detection module (104) to notify the administrator of any updates made to a user's biometric registration based on the detected changes and a confidence score. system after claim 1 wherein the multimodal biometric authentication module (102) is configured to: capture the pair of biometrics with at least high consistency, such as B. an iris, or low permanence, such. B. the signature. extracting at least one feature vector from the detected pair of biometric features by a feature extraction module (102a); And classify the extracted feature vectors by a multimodal classification module (102b) to determine a user's identity over a threshold period and confidence level based on a convolutional neural network and long-term short-term memory (LSTM). system after claim 3 , wherein the extracted features are combined together to form a pair of unimodal modules for each pair of biometric features. system after claim 3 , wherein an initialization module (112) is connected to the feature extraction module (102a) to define a maximum initial low confidence threshold as an average confidence value resulting from the validation of the input and to set a threshold time of updating the feature vector of a variant biometric feature for a period of more than 6 months from a current date. system after claim 5 , wherein a comparison module (114) is connected to the initialization module (112) and the multi-modal biometric authentication module (102) to compare the obtained confidence with the initialized confidence value and the obtained time with the threshold time. system after claim 1 , wherein an update module (116) is connected to the comparison module (114) and the blockchain network (108) to trigger a smart contract to update the feature vector, the update module (116) being triggered when the confidence is below the threshold and the threshold time is either greater than or equal to the initialized threshold time.

Images