# US 11,444,774 B2 · Method and system for biometric verification

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 11,444,774 B2
- Document ID: US11444774B2
- Status: Granted
- Kind: B2
- Application number: 17/026,380
- Filed: 2020-09-21
- Published: 2021-07-08
- Granted: 2022-09-13
- Earliest claimed: 2020-01-08
- Assignees: Tata Consultancy Services Limited
- Applicants: Tata Consultancy Services Limited
- Inventors: Arun Kumar Jindal; Vasudha Kumari; Imtiyazuddin Shaik; Srinivasa Rao Chalamala; Rajan Mindigal Alasingara Bhattachar; Sachin Premsukh Lodha
- Web page: https://patentgrep.com/patent/US11444774B2

## CPC

- H04L 9/3231: {Biological data, e.g. fingerprint, voice or retina (network architectures or network communication protocols for supporting authentication of entities using biometrical features in a packet data network H04L63/0861)}
- H04L 9/008: {involving homomorphic encryption}
- H04L 9/0897: {involving additional devices, e.g. trusted platform module [TPM], smartcard or USB}
- H04L 9/3006: {underlying computational problems or public-key parameters}
- H04L 9/3263: {involving certificates, e.g. public key certificate [PKC] or attribute certificate [AC]; Public key infrastructure [PKI] arrangements (network architectures or network communication protocols for supporting authentication of entities using certificates in a packet data network H04L63/0823)}
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

This disclosure relates generally to a method and system for biometric verification. Conventional biometric verification method and system performs one or more computations in non-encrypted domain, thereby leading to security threats. The disclosed method includes performing computations such as enrollment and verification feature vector computation, dimensionality reduction of said feature vectors, and comparison of dimensionally reduced encrypted feature vectors to obtain matching scores indicating the extent of match therebetween between in encrypted domain using fully homomorphic encryption, thereby leading to secure biometric verification.

## Claims

### Claim 1

A processor-implemented method for biometric verification, comprising:
acquiring, via one or more hardware processors, a first biometric sample, the first biometric sample comprising at least a portion of a first high-resolution image of a biometric modality of a user;
performing, via the one or more hardware processors, data augmentation on at least the portion of the first high-resolution image to obtain a set of augmented image portions of the first biometric sample;
obtaining a set of first feature vectors corresponding to the set of augmented image portions and the first high resolution image, via the one or more hardware processors;
encrypting, using Fully Homomorphic Encryption, each first feature vector of the set of first feature vectors using a public key stored at a first computation device to obtain a set of first encrypted feature vectors, via the one or more hardware processors;
obtaining, via the one or more hardware processors, a set of encrypted reduced dimensionality first feature vectors corresponding to the set of encrypted first feature vectors, wherein an encrypted reduced dimensionality first feature vector from amongst the set of encrypted reduced dimensionality first feature vectors corresponding to an encrypted first feature vector from amongst the set of encrypted first feature vector is obtained by performing a homomorphic operation of the encrypted feature vector with an encrypted random projection matrix encrypted using Fully Homomorphic Encryption and, and wherein the encrypted random projection matrix is pre-assigned to the user and is encrypted using the public key;
sharing, via the one or more hardware processors, the set of encrypted reduced dimensionality first feature vectors with a second computation device, wherein the set of encrypted reduced dimensionality feature vectors are compared with a pre-stored homomorphic encrypted second feature vector corresponding to a second biometric sample associated with the user to obtain a set of matching scores, wherein a matching score between an encrypted reduced dimensionality first feature vector of the set of encrypted reduced dimensionality first feature vectors and the homomorphic encrypted second feature vector is indicative of an extent of matching between the first biometric sample and the second biometric sample, and wherein each matching score of the set of encrypted matching scores is encrypted via homomorphic encryption using the public key;
receiving, from the second device, the set of encrypted matching scores and decrypting the set of encrypted matching scores using a private key, via the one or more hardware processors; and
verifying the user based on a comparison of the set of matching scores with a predetermined threshold score, via the one or more hardware processors.

### Claim 2 (depends on 1)

The method of claim 1, wherein acquiring the first biometric sample comprises:
capturing the first high resolution image of at least one biometric modality of the user; and
segmenting the first high resolution image to obtain the first biometric sample.

### Claim 3 (depends on 1)

The method of claim 1, wherein performing the data augmentation on at least the portion of the first high-resolution image comprises performing one or more operations on at least the portion of the first high-resolution image, the one or more operations comprises randomly zooming, shifting horizontally, shifting vertically, flipping horizontally, randomly rotating, changing brightness, cropping and resizing.

### Claim 4 (depends on 1)

The method of claim 1, wherein the pre-stored homomorphic encrypted second feature vector corresponding to the second biometric sample associated with the user is stored by:
acquiring the second biometric sample comprising at least the portion of a second high-resolution image of the biometric modality of the user during an enrollment phase;
obtaining the second feature vector corresponding to the at least the portion of the second high-resolution image;
encrypting, using Fully Homomorphic Encryption, the second feature vector using the public key stored at the first computation device to obtain an encrypted second feature vector;
obtaining an encrypted reduced dimensionality second feature vector corresponding to the encrypted second feature vector, wherein the encrypted reduced dimensionality second feature vector is obtained by performing a homomorphic operation of the encrypted second feature vector with an encrypted random projection matrix encrypted using Fully Homomorphic Encryption, the encrypted random projection matrix pre-assigned to the user, and wherein the encrypted random projection matrix is encrypted using the public key; and
sharing the encrypted reduced dimensionality second feature vector with the second computation device.

### Claim 5 (depends on 1)

The method of claim 1, wherein obtaining the matching score between the encrypted reduced dimensionality first feature vector and the homomorphic encrypted second feature vector comprises:
packing the encrypted reduced dimensionality first feature vector in a first ciphertext and the homomorphic encrypted second feature vector in a second ciphertext;
multiplying, elementwise, the first ciphertext with the second ciphertext to obtain a resultant vector;
rotating the resultant vector log(n) times to obtain a plurality of rotated resultant vectors; and
computing a sum of the plurality of rotated resultant vectors to obtain an inner product of the encrypted reduced dimensionality first feature vector and the homomorphic encrypted second feature vector, wherein the inner product represents the matching score between the encrypted reduced dimensionality first feature vector and the homomorphic encrypted second feature vector.

### Claim 6 (depends on 1)

The method of claim 1, wherein the first high-resolution image is acquired during a verification phase of the biometric verification.

### Claim 7 (depends on 1)

The method of claim 1, further comprising sharing, with the second device, the public key, encrypted random projection matrix, and a unique identifier (ID) associated with the user.

### Claim 8 (depends on 1)

The method of claim 1, wherein the public key and the private key are associated with the user.

### Claim 9 (depends on 1)

The method of claim 1, wherein the first computation device is a client device and the second computation device is a server device.

### Claim 10

A system for biometric verification comprising:
one or more memories; and
one or more hardware processors, the one or more memories coupled to the one or more hardware processors, wherein the one or more hardware processors are configured to execute programmed instructions in a trusted execution environment (TEE), the programmed instructions stored in the one or more memories, to:
acquire a first biometric sample, the first biometric sample comprising at least a portion of a first high-resolution image of a biometric modality of a user;
perform data augmentation on at least the portion of the first high resolution image to obtain a set of augmented image portions of the first biometric sample;
obtain a set of first feature vectors corresponding to the set of augmented image portions and the first high resolution image;
encrypt, using Fully Homomorphic Encryption, each first feature vector of the set of first feature vectors using a public key stored at a first computation device to obtain a set of first encrypted feature vectors;
obtain a set of encrypted reduced dimensionality first feature vectors corresponding to the set of encrypted first feature vectors, wherein an encrypted reduced dimensionality first feature vector from amongst the set of encrypted reduced dimensionality first feature vectors corresponding to an encrypted first feature vector from amongst the set of encrypted first feature vector is obtained by performing a homomorphic operation of the encrypted feature vector with an encrypted random projection matrix encrypted using Fully Homomorphic Encryption, and wherein the encrypted random projection matrix is pre-assigned to the user and encrypted using the public key;
share the set of encrypted reduced dimensionality first feature vectors with a second computation device, wherein the set of encrypted reduced dimensionality feature vectors are compared with a pre-stored homomorphic encrypted second feature vector corresponding to a second biometric sample associated with the user to obtain a set of matching scores, wherein a matching score between an encrypted reduced dimensionality first feature vector of the set of encrypted reduced dimensionality first feature vectors and the homomorphic encrypted second feature vector is indicative of an extent of matching between the first biometric sample and the second biometric sample, and wherein each matching score of the set of encrypted matching scores is encrypted via homomorphic encryption using the public key;
receive, from the second device, the set of encrypted matching scores and decrypting the set of encrypted matching scores using a private key; and
verify the user based on a comparison of the set of matching scores with a predetermined threshold score.

### Claim 11 (depends on 10)

The system of claim 10, wherein to acquire the first biometric sample, the one or more hardware processors are configured by the instructions to:
capture the first high resolution image of at least one biometric modality of the user; and
segment the first high resolution image to obtain the first biometric sample.

### Claim 12 (depends on 11)

The system of claim 11, wherein the public key and the private key are associated with the user.

### Claim 13 (depends on 11)

The system of claim 11, wherein the first computation device is a client device.

### Claim 14 (depends on 11)

The system of claim 11, wherein the second computation device is a server device.

### Claim 15 (depends on 10)

The system of claim 10, wherein to perform the data augmentation on the at least portion of the first high resolution image, the one or more hardware processors are configured by the instructions to perform one or more operations on at least the portion of the first high resolution image, the one or more operations comprises randomly zooming, shifting horizontally, shifting vertically, flipping horizontally, randomly rotating, changing brightness, cropping and resizing.

### Claim 16 (depends on 10)

The system of claim 10, wherein to store the pre-stored homomorphic encrypted second feature vector corresponding to the second biometric sample associated with the user, the one or more hardware processors are configured by the instructions to:
acquire the second biometric sample comprising at least the portion of a second high-resolution image of the biometric modality of the user during an enrollment phase;
obtain the second feature vector corresponding to the at least the portion of the second high-resolution image;
encrypt, using Fully Homomorphic Encryption, the second feature vector using the public key stored at the first computation device to obtain an encrypted second feature vector;
obtain an encrypted reduced dimensionality second feature vector corresponding to the encrypted second feature vector, wherein the encrypted reduced dimensionality second feature vector is obtained by performing a homomorphic operation of the encrypted second feature vector with an encrypted random projection matrix encrypted using Fully Homomorphic Encryption, the encrypted random projection matrix pre-assigned to the user, and wherein the encrypted random projection matrix is encrypted using the public key; and
share the encrypted reduced dimensionality second feature vector with the second computation device.

### Claim 17 (depends on 10)

The system of claim 10, wherein to obtain the matching score between the encrypted reduced dimensionality first feature vector and the homomorphic encrypted second feature vector, the one or more hardware processors are configured by the instructions to:
pack the encrypted reduced dimensionality first feature vector in a first ciphertext and the homomorphic encrypted second feature vector in a second ciphertext;
multiply, elementwise, the first ciphertext with the second ciphertext to obtain a resultant vector;
rotate the resultant vector log(n) times to obtain a plurality of rotated resultant vectors; and
compute a sum of the plurality of rotated resultant vectors to obtain an inner product of the encrypted reduced dimensionality first feature vector and the homomorphic encrypted second feature vector, wherein the inner product represents the matching score between the encrypted reduced dimensionality first feature vector and the homomorphic encrypted second feature vector.

### Claim 18 (depends on 10)

The system of claim 10, wherein the first high resolution image is acquired during a verification phase of the biometric verification.

### Claim 19 (depends on 10)

The system of claim 10, wherein the one or more hardware processors are
further configured by the instructions to share, with the second device, the public key, encrypted random projection matrix, and a unique identifier (ID) associated with the user.

### Claim 20

One or more non-transitory machine readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause:
acquiring, via one or more hardware processors, a first biometric sample, the first biometric sample comprising at least a portion of a first high-resolution image of a biometric modality of a user;
performing, via the one or more hardware processors, data augmentation on at least the portion of the first high-resolution image to obtain a set of augmented image portions of the first biometric sample;
obtaining a set of first feature vectors corresponding to the set of augmented image portions and the first high resolution image, via the one or more hardware processors;
encrypting, using Fully Homomorphic Encryption, each first feature vector of the set of first feature vectors using a public key stored at a first computation device to obtain a set of first encrypted feature vectors, via the one or more hardware processors;
obtaining, via the one or more hardware processors, a set of encrypted reduced dimensionality first feature vectors corresponding to the set of encrypted first feature vectors, wherein an encrypted reduced dimensionality first feature vector from amongst the set of encrypted reduced dimensionality first feature vectors corresponding to an encrypted first feature vector from amongst the set of encrypted first feature vector is obtained by performing a homomorphic operation of the encrypted feature vector with an encrypted random projection matrix encrypted using Fully Homomorphic Encryption and, and wherein the encrypted random projection matrix is pre-assigned to the user and is encrypted using the public key;
sharing, via the one or more hardware processors, the set of encrypted reduced dimensionality first feature vectors with a second computation device, wherein the set of encrypted reduced dimensionality feature vectors are compared with a pre-stored homomorphic encrypted second feature vector corresponding to a second biometric sample associated with the user to obtain a set of matching scores, wherein a matching score between an encrypted reduced dimensionality first feature vector of the set of encrypted reduced dimensionality first feature vectors and the homomorphic encrypted second feature vector is indicative of an extent of matching between the first biometric sample and the second biometric sample, and wherein each matching score of the set of encrypted matching scores is encrypted via homomorphic encryption using the public key;
receiving, from the second device, the set of encrypted matching scores and decrypting the set of encrypted matching scores using a private key, via the one or more hardware processors; and
verifying the user based on a comparison of the set of matching scores with a predetermined threshold score, via the one or more hardware processors.