# US 2019/0244390 A1 · Face authentication method

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

- Number: US 2019/0244390 A1
- Document ID: US20190244390A1
- Status: Published application
- Kind: A1
- Application number: 16/267,110
- Filed: 2019-02-04
- Published: 2019-08-08
- Earliest claimed: 2018-02-06
- Assignees: Idemia Identity & Security France
- Applicants: Idemia Identity & Security France
- Inventors: Julien DOUBLET; Jean BEAUDET
- Web page: https://patentgrep.com/patent/US20190244390A1

## CPC

- G06T 7/74: {involving reference images or patches}
- G06F 21/32: using biometric data, e.g. fingerprints, iris scans or voiceprints
- G06V 40/172: {Classification, e.g. identification}
- G06T 2207/30201: Face
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

The invention relates to a method for authenticating a face presented to a device comprising the steps in which the imager acquires an initial image of a face, the processing unit determines a pose of the face from the initial image, and determines a reference pose and a target position placed randomly or pseudo-randomly in a target space, the screen displays a displayed image (6) comprising at least one visual orientation mark (7) initially at a reference position, and a visual target (8) at the target position, updated by moving the visual orientation mark (7) according to the successive poses of the face, and the processing unit authenticates the face presented if there is a match between the position of the visual orientation mark (7) and the target position where the visual target is located (8).

## Claims

### Claim 1

A method for authenticating a face presented to a device comprising an imager, a screen and a processing unit, comprising the steps in which:
the imager acquires at least one initial image of a face presented in the acquisition field of the imager,
the processing unit determines a pose of the face from the initial image, and from this face pose, determines a reference pose, a face pose being defined by at least two pose angles representative of an orientation of the face appearing in an image, and the processing unit determines a target position placed randomly or pseudo-randomly in a target space at a distance from a reference position corresponding to the reference pose, said target position corresponding to a target pose,
the screen displays a displayed image comprising at least one visual orientation mark initially at the reference position determined from the pose of the face in the initial image, and a visual target at the target position,
the imager acquires a stream of images of the face presented in the acquisition field, and for a plurality of successive images of said image stream:
the processing unit determines a pose of the face in each image of the plurality of successive images,
the processing unit determines successive positions of the visual orientation mark according to the successive poses of the face with respect to the reference pose in the plurality of successive images,
the image displayed by the screen is updated by moving the visual orientation mark according to the successive positions of the visual orientation mark determined according to the successive poses of the face with respect to the reference pose in the plurality of successive images,
the processing unit authenticates the presented face when, following the displacement of the visual orientation mark, there is a match between the position of the visual orientation mark and the target position where the visual target is located, corresponding to a match between the face pose and the target pose, otherwise the face presented in the acquisition field is considered as fraud.

### Claim 2 (depends on 1)

The method according to claim 1, wherein the target space extends on either side of the reference position.

### Claim 3 (depends on 1)

The method according to claim 1, wherein the target space corresponds to angular ranges of the pose angles defining a plurality of possible target positions.

### Claim 4 (depends on 3)

The method according to claim 3, wherein the angles representative of the orientation of the face appearing in the image comprise a yaw angle about a vertical axis and a pitch angle about a horizontal axis, the angular ranges extending between ±10° and ±20° with respect to the pose angles of the reference pose.

### Claim 5 (depends on 1)

The method according to claim 1, wherein the processing unit defines the target space also according to user data and/or elements contained in the initial image.

### Claim 6 (depends on 1)

The method according to claim 1, wherein a time is allotted at the expiry of which, in the absence of match between the position of the visual orientation mark and the target position where the visual target is located, the face presented in the acquisition field is considered as fraud.

### Claim 7 (depends on 1)

The method according to claim 1, wherein the match between the position of the visual orientation mark and the target position where the visual target is located, needs to be maintained for a predetermined duration before the processing unit authenticates the face.

### Claim 8 (depends on 1)

The method according to claim 1, wherein, when the processing unit determines a reference pose, the processing unit transmits an order of movement of the face if the face pose does not correspond to a front view of the face.

### Claim 9 (depends on 1)

The method according to claim 1, wherein the authentication of the presented face is also conditioned by the determination of the three-dimensional appearance of the presented face by implementing a photogrammetric reconstruction technique based on a movement from at least two face images of the image stream corresponding to two different poses.

### Claim 10 (depends on 1)

The method according to claim 1, wherein the processing unit determines, for the plurality of successive images of the image stream, a direction of the face gaze by identifying an orientation of the eyes of said face on the acquired images, the authentication of the presented face is also conditioned on a gaze in a direction of the visual orientation mark and/or the visual target.

### Claim 11 (depends on 1)

A computer-readable non-transitory medium comprising program code instructions stored thereon for performing the steps of a method according to claim 1 when said program is executed on a computer reading said non-transitory medium.

### Claim 12 (depends on 1)

A device comprising an imager, a screen and a processing unit, said device being configured to implement a method according to claim 1.
