# US 11,909,854 B2 · Third party biometric homomorphic encryption matching for privacy protection

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

- Number: US 11,909,854 B2
- Document ID: US11909854B2
- Status: Granted
- Kind: B2
- Application number: 18/080,310
- Filed: 2022-12-13
- Published: 2023-12-14
- Granted: 2024-02-20
- Earliest claimed: 2022-06-09
- Assignees: The Government of the United States of America, as represented by the Secretary of Homeland Security
- Applicants: The Government of the United States of America, as represented by the Secretary of Homeland Security
- Inventors: Arun Vemury
- Web page: https://patentgrep.com/patent/US11909854B2

## CPC

- H04L 9/008: {involving homomorphic encryption}
- H04L 9/0894: {Escrow, recovery or storing of secret information, e.g. secret key escrow or cryptographic key storage}
- 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: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

Systems and methods for secure distribution of biometric matching processing are provided. Certain configurations include homomorphic encrypting of captured biometric information. In some configurations, the biometric information is classified without decryption between a first identity class and a second identity class. The biometric information may be formed as a feature vector. A homomorphic encrypted feature vector may be formed by homomorphic encrypting of the biometric information.

## Claims

### Claim 1

A system comprising:
a reference gallery of homomorphic encrypted (HE) biometric information;
a third-party computer;
HE biometric matching logic;
an access control device;
a kiosk comprising a scanner connected to the reference gallery via a cloud resource; the scanner configured to:
generate biometric information to store in the reference gallery;
scan a user and generate biometric information identifying the user from the scan;
generate HE encrypted information containing biometric data using an HE encryption algorithm;
the system configured to:
store the generated HE encrypted biometric information in a library;
distribute the HE encrypted biometric information to the third-party computer;
an HE encryption logic configured to construct, based on the reference gallery, HE encrypted reference information;
the HE encryption logic configured to store HE encrypted information in the reference gallery;
the HE encryption logic configured to apply homomorphic encryption to information captured from the biometric capture device;
the third-party computer comprising HE biometric matching logic configured to perform a HE biometric matching process;
the HE biometric matching logic configured to perform a matching comparison of the HE encrypted information from the library against HE encrypted biometric information in the reference gallery to identify matching HE encrypted reference information;
the HE biometric matching logic configured to perform the HE biometric matching process without decrypting or having a capacity to decrypt the HE encrypted information;
the third-party computer configured to process biometric information used for biometric matching, without access to underlying biometric information of the user;
the third-party computer configured to transmit a message to the access control device;
the access control device programmed to execute an access granted process when the message indicates a positive match based on the matching comparison; and
the access control device programmed to execute an access denied process when the message does not indicate a positive match based on the matching comparison.

### Claim 2 (depends on 1)

The system of claim 1 wherein the biometric reference gallery is configured to store for each of one or more of N identities, a plurality of reference biometric images including gait or other movement biometrics; wherein N is selected from the group of natural numbers.

### Claim 3 (depends on 2)

The system of claim 2 wherein the reference biometric images include a visible image pixel array.

### Claim 4 (depends on 3)

The system of claim 3 wherein the system is configured to obtain information characterizing features of the reference biometric images by extracting information from a captured image.

### Claim 5 (depends on 4)

The system of claim 4 wherein the reference gallery is configured to store the characterizing features as an image feature vector.

### Claim 6 (depends on 5)

The system of claim 5 wherein the image feature vector includes a principal component analysis (PCA) having an orthogonal linear transformation to transform data into a new coordinate system such that a first greatest variance by scalar projections of related data will lie on a first principal component, and a second greatest variance on a second principal component.

### Claim 7 (depends on 5)

The system of claim 5 wherein the image feature vector comprises a reduced dimensionality to conserve memory space.

### Claim 8 (depends on 2)

The system of claim 2 comprising:
an uploading station comprising a biometric capture device; and
HE encryption logic configured to apply homomorphic encryption to information captured from the biometric capture device.

### Claim 9 (depends on 8)

The system of claim 8 configured to perform HE domain classification on received biometric information from the uploading station.

### Claim 10 (depends on 2)

The system of claim 2 comprising a classifier configured to select a type of HE encryption to program the third-party computer.

### Claim 11 (depends on 2)

The system of claim 2 comprising an uploading station configured to upload encrypted biometric information to the third-party computer; wherein said third-party computer is configured to receive a HE encrypted biometric descriptor from the uploading station.

### Claim 12 (depends on 11)

The system of claim 11 comprising a classifier configured to select a type of HE encryption to program the third-party computer; wherein the classifier is configured to classify the HE encrypted biometric descriptor among N identity classes, without decryption of the HE encrypted biometric descriptor, and without access to the N identities.

### Claim 13 (depends on 12)

The third-party computer of claim 12 further configured to receive a distribution of HE encrypted information; wherein the classifier is configured to classify the distribution among the N identity classes without the third-party computer configured to obtain access to the N identities and without the third-party computer having access to the N identities.

### Claim 14 (depends on 12)

The classifier of claim 12 comprising HE domain third party classifier configuration logic.

### Claim 15 (depends on 14)

The classifier of claim 14 configured to:
construct an HE mode N class classifier; and
communicate the HE mode N class classifier to the third-party computer via a network or cloud service; communication of the HE mode N class classifier including an initial communication of an initial N class classifier.

### Claim 16 (depends on 15)

The system of claim 15 configured to:
receive the HE mode N class classifier constructed by the HE domain third party classifier configuration logic; and
with the HE mode N class classifier, the third-party computer configured to load or install the HE mode N class classifier as a loaded HE domain classifier classification.

### Claim 17 (depends on 16)

The system of claim 16 wherein:
the third-party computer comprises classification output logic;
the classification output logic configured to communicate results from the loaded HE domain classifier to an uploading station;
the uploading station configured to upload the HE encrypted biometric descriptor to the third-party computer; and
the uploading station configured to receive the loaded HE domain classifier classification of the HE encrypted biometric descriptor; the uploading station configured to display results from the loaded HE domain classifier classification of the HE encrypted biometric descriptor.

### Claim 18 (depends on 16)

The classifier of claim 16 wherein the HE domain third party classifier configuration logic is configured to:
first construct a clear-text version of the HE mode, N-class CNN biometric classification model;
followed by training the clear-text version using N reference images stored in the reference gallery;
after the training, the HE domain third party classifier configuration logic configured to convert the clear-text version of the HE mode, N-class CNN biometric classification model to the HE mode N Class classifier; and
the HE domain third party classifier configuration logic configured to then upload the HE mode, N-class CNN biometric classification model to the third-party computer.

### Claim 19 (depends on 12)

The classifier of claim 12 configured to:
use fully homomorphic encryption; and
construct a HE domain artificial neural network for uploading to the third-party computer; the HE domain artificial neural network comprising an HE domain N-class convolutional neural network (CNN) biometric classification model.

### Claim 20

A method comprising:
storing homomorphic encrypted (HE) biometric information in a reference gallery;
providing HE biometric matching logic, and an access control device;
generating biometric information to store in the reference gallery with a kiosk comprising a scanner;
scanning a user and generating biometric information for identifying the user;
generating HE encrypted information containing biometric data using an HE encryption algorithm;
storing the generated HE encrypted biometric information in a library;
distributing the HE encrypted biometric information to a third-party computer; the third-party computer comprising HE Biometric matching logic;
based on the reference gallery, constructing HE encrypted reference information using HE encryption logic;
the HE Encryption logic storing HE encrypted information in the reference gallery;
the third-party computer performing a HE biometric matching process using the HE biometric matching logic;
the HE biometric matching logic performing a matching comparison of the HE encrypted information from the library against HE encrypted biometric information in the reference gallery to identify matching HE encrypted reference information;
the HE biometric matching logic performing the HE biometric matching process without decrypting or having a capacity to decrypt the HE encrypted information;
the third-party computer processing biometric information used for biometric matching, without access to underlying biometric information of the user;
the third-party computer transmitting a message to the access control device;
the access control device executing an access granted process when the message indicates a positive match based on the matching comparison; and
the access control device executing an access denied process when the message does not indicate a positive match based on the matching comparison.

### Claim 21 (depends on 19)

The method of claim 19 comprising:
a classifier selecting a type of HE encryption to program the third-party computer; and
the classifier classifying the HE encrypted biometric descriptor among N identity classes, without decryption of the HE encrypted biometric descriptor, and without access to the N identities.

### Claim 22 (depends on 21)

The method of claim 21 comprising:
receiving a distribution of HE encrypted information; and
the classifier classifying the distribution among the N identity classes without the third-party computer configured to obtain access to the N identities and without the third-party computer having access to the N identities.

### Claim 23 (depends on 21)

The classifier of claim 21 comprising HE domain third party classifier configuration logic.

### Claim 24 (depends on 23)

The method of claim 23 comprising:
the classifier constructing an HE mode N class classifier;
communicating the HE mode N class classifier to the third-party computer via a network or cloud service; communication of the HE mode N class classifier including an initial communication of an initial N class classifier.

### Claim 25 (depends on 24)

The method of claim 24 comprising:
receiving the HE mode N class classifier constructed by the HE domain third party classifier configuration logic; and
with the HE mode N class classifier, the third-party computer loading or installing the HE mode N class classifier as a loaded HE domain classifier classification.

### Claim 26 (depends on 25)

The method of claim 25 comprising:
classification output logic communicating results from the loaded HE domain classifier to an uploading station;
the uploading station uploading an HE encrypted biometric descriptor to the third-party computer; and
the uploading station receiving the loaded HE domain classifier classification of the HE encrypted biometric descriptor; and
the uploading station displaying results from the loaded HE domain classifier classification of the HE encrypted biometric descriptor.

### Claim 27 (depends on 25)

The method of claim 25 comprising:
the HE domain third party classifier configuration logic first constructing a clear-text version of the HE mode, N-class CNN biometric classification model;
followed by training the clear-text version using N reference images stored in the reference gallery;
after the training, the HE domain third party classifier configuration logic converting the clear-text version of the HE mode, N-class CNN biometric classification model to the HE mode N Class classifier; and
the HE domain third party classifier configuration logic uploading the HE mode, N-class CNN biometric classification model to the third-party computer.

### Claim 28 (depends on 21)

The method of claim 21 comprising the classifier using fully homomorphic encryption.

### Claim 29 (depends on 21)

The method of claim 21 construct a HE domain artificial neural network for uploading to the third-party computer; the HE domain artificial neural network comprising an HE domain N-class convolutional neural network (CNN) biometric classification model.