Facial liveness detection in image biometrics

• US 9,672,415 B2
• Filed: 03/27/2015
• Issued: 06/06/2017
• Est. Priority Date: 11/13/2014
• Status: Active Grant

• Current Assignee: Tahoe Research Limited (f/k/a Learndale Limited) (Vector Capital Corporation)
• Original Assignee: Intel Corporation
• Inventors: Ghosh, Rahuldeva, Negi, Ansuya, Pfursich, Scott, Graumann, David L., Narjala, Ranjit S
• Primary Examiner(s): Park, Edward
• Application Number:
  US14/779,540
• Publication Number:
  US 20160335483A1
• Time in Patent Office: 802 Days
• CPC Class Codes: G06V 10/40 Extraction of image or video features
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation
• CPC Class Codes: G06V 40/167 using comparisons between temporally consecutive images
• CPC Class Codes: G06V 40/168 Feature extraction; Face representation
• CPC Class Codes: G06V 40/172 Classification, e.g. identification
• CPC Class Codes: G06V 40/174 Facial expression recognition
• CPC Class Codes: G06V 40/19 Sensors therefor
• CPC Class Codes: G06V 40/40 Spoof detection, e.g. liveness detection
• CPC Class Codes: G06V 40/70 Multimodal biometrics, e.g. combining information from different biometric modalities
• CPC Class Codes: H04L 63/0861 using biometrical features, e.g. fingerprint, retina-scan cryptographic mechanisms or cryptographic arrangements for entity authentication using biological data H04L9/3231
• CPC Class Codes: H04W 12/065 Continuous authentication

Abstract

System and techniques for spoofing detection in image biometrics are described herein. A sequence of images may be obtained from a camera; a first plurality of images in the sequence of images including a representation of a user body part, and a second plurality of images in the sequence of images including a representation of an environment of the user. A marker may be created for the representation of the body part. A feature of the environment of the user present during the second plurality of images may be identified in the sequence of images using a third group of circuits. A correlation between the marker and the feature of the environment in the sequence of images may be quantified to produce a synchronicity metric of the degree to which the marker and the feature of the environment correlate.

Claims

1. A system for facial liveness detection in image biometrics, the system comprising:

a biometric feature detector to detecting a set of potential authentication faces in an image sequence captured during an authentication attempt by a user;

a liveness indication controller to;

for each face in the set of potential authentication faces;

use an emitter of the system to emit light towards the face;

use a sensor of the system to measure an interaction of the light with the face; and

invoke a synchronicity detector to;

apply a set of liveness model tests to the face, each test in the set of liveness tests including an assay that differs between a live face and a face simulated on a pictorial representation of a face, the assay operating independently of body arrangement instructions given to the user; and

compare the interaction of the light with the face with a model of a live face, wherein in at least one test in the set of liveness models tests, the interaction of the light with the face provides a distance measurement between the face and the emitter, wherein the model of the live face includes an acceptable size range of a live face, and wherein to compare the interaction with the model includes the synchronicity detector to scale a visual light spectrum representation of the face acquired from the image sequence using the distance measurement to determine whether the face is within the acceptable size range of a live face; and

add the face to a set of authentication faces when the result of the set of liveness model tests meets a threshold; and

an authentication controller to apply facial authentication to faces in the set of authentication faces.

2. The system of claim 1, wherein the interaction of the light with the face is a reflectivity measurement of the character of the light reflected from the face.

3. The system of claim 2, wherein to emit the light includes the emitter to emit non-visual spectrum light, wherein the model of a live face includes a non-uniform positive reflectivity of the light, and wherein to compare the interaction with the model of the face includes the synchronicity detector to determine whether the area in which the face was found has non-uniform positive reflectivity in the non-visual spectrum light.

4. The system of claim 2, wherein to emit the light include the emitter to emit a non-uniform pattern of visible light, wherein the model of a live face does not reflect the pattern, and wherein to compare the interaction with the model of the face includes the synchronicity detector to determine whether the pattern is discernable over the face.

5. The system of claim 1, wherein another test in the liveness model of tests includes providing a target, without an instruction to the user, on a display.

6. The system of claim 5, wherein the target is moving, wherein the test from the set of liveness model tests includes the biometric feature detector to track an eye gaze of the user with respect to the target, and wherein the assay is an evaluation of how closely the eye gaze tracks the target.

7. The system of claim 6, wherein the target is a frame superimposed on the sequence of images re-displayed to the user via a display of the system, wherein the test from the set of liveness model tests includes the synchronicity detector to determine a synchronous movement between the user and an environmental feature ascertainable from the sequence of images, and wherein the assay is an evaluation of how closely movement of the environmental feature is synchronized to movement of the user.

8. The system of claim 6, wherein the target is an icon in a display area including an avatar of the user, the display area not including the sequence of images by an environmental feature detector, wherein the test from the set of liveness model tests includes the synchronicity detector to determine a synchronous movement between the user and an environmental feature ascertainable from the sequence of images, and wherein the assay is an evaluation of how closely movement of the environmental feature is synchronized to movement of the user.

9. The system of claim 1, wherein another test in the liveness model of tests includes the environmental feature detector identifying an outline conforming to a device across the sequence of images, and wherein the assay is whether the face appears within the outline across a plurality of the sequence of images.

10. The system of claim 9, wherein whether the face appears within the outline across the plurality of the sequence of images is true if the face appears in an area of an image in the sequence of images and the outline encloses the area in a set of preceding images in the sequence of images,

the set of preceding images selected based on a movement threshold applied to the outline.

11. A method for facial liveness detection in image biometrics, the method comprising:

detecting a set of potential authentication faces in an image sequence captured during an authentication attempt by a user;

for each face in the set of potential authentication faces, perform the operations of;

emitting light towards the face with an emitter;

measuring an interaction of the light with the face;

comparing the interaction with a model of a live face; and

applying a set of liveness model tests to the face, each test in the set of liveness tests including an assay that differs between a live face and a face simulated on a pictorial representation of a face, the assay operating independently of body arrangement instructions given to the user, wherein in at least one test in the set of liveness models tests, the interaction of the light with the face provides a distance measurement between the face and the emitter, wherein the model of the live face includes an acceptable size range of a live face, and wherein to compare the interaction with the model includes the synchronicity detector to scale a visual light spectrum representation of the face acquired from the image sequence using the distance measurement to determine whether the face is within the acceptable size range of a live face; and

adding the face to a set of authentication faces when the result of the set of liveness model tests meets a threshold; and

applying facial authentication to faces in the set of authentication faces.

12. The method of claim 11, wherein the interaction of the light with the face is a reflectivity measurement of the character of the light reflected from the face.

13. The method of claim 12, wherein emitting the light includes emitting non-visual spectrum light, wherein the model of a live face includes a non-uniform positive reflectivity of the light, and wherein comparing the interaction with the model of the face includes determining whether the area in which the face was found has non-uniform positive reflectivity in the non-visual spectrum light.

14. The method of claim 12, wherein emitting the light include emitting a non-uniform pattern of visible light, wherein the model of a live face does not reflect the pattern, and wherein comparing the interaction with the model of the face includes determining whether the pattern is discernable over the face.

15. The method of claim 11, wherein another test in the liveness model of tests includes providing a target, without an instruction to the user, on a display.

16. The method of claim 15, wherein the target is moving, wherein the test from the set of liveness model tests includes tracking an eye gaze of the user with respect to the target, and wherein the assay is an evaluation of how closely the eye gaze tracks the target.

17. The method of claim 15, wherein the target is a frame superimposed on the sequence of images re-displayed to the user, wherein the test from the set of liveness model tests includes determining a synchronous movement between the user and an environmental feature ascertainable from the sequence of images, and wherein the assay is an evaluation of how closely movement of the environmental feature is synchronized to movement of the user.

18. The method of claim 15, wherein the target is an icon in a display area including an avatar of the user, the display area not including the sequence of images, wherein the test from the set of liveness model tests includes determining a synchronous movement between the user and an environmental feature ascertainable from the sequence of images, and wherein the assay is an evaluation of how closely movement of the environmental feature is synchronized to movement of the user.

19. The method of claim 11, wherein another test in the liveness model of tests includes identifying an outline conforming to a device across the sequence of images, and wherein the assay is whether the face appears within the outline across a plurality of the sequence of images.

20. At least one non-transitory machine readable medium including instructions for facial liveness detection in image biometrics, the instructions, when executed by a machine, cause the machine to perform operations comprising:

detecting a set of potential authentication faces in an image sequence captured during an authentication attempt by a user;

for each face in the set of potential authentication faces, perform the operations of;

emitting light towards the face with an emitter;

measuring an interaction of the light with the face;

comparing the interaction with a model of a live face; and

applying a set of liveness model tests to the face, each test in the set of liveness tests including an assay that differs between a live face and a face simulated on a pictorial representation of a face, the assay operating independently of body arrangement instructions given to the user, wherein in at least one test in the set of liveness models tests, the interaction of the light with the face provides a distance measurement between the face and the emitter, wherein the model of the live face includes an acceptable size range of a live face, and wherein to compare the interaction with the model includes the synchronicity detector to scale a visual light spectrum representation of the face acquired from the image sequence using the distance measurement to determine whether the face is within the acceptable size range of a live face;

adding the face to a set of authentication faces when the result of the set of liveness model tests meets a threshold; and

applying facial authentication to faces in the set of authentication faces.

21. The at least one non-transitory machine readable medium of claim 20, wherein the interaction of the light with the face is a reflectivity measurement of the character of the light reflected from the face.

22. The at least one non-transitory machine readable medium of claim 20, wherein another test in the liveness model of tests includes providing a target, without an instruction to the user, on a display.

23. The at least one non-transitory machine readable medium of claim 20, wherein another test in the liveness model of tests includes identifying an outline conforming to a device across the sequence of images, and wherein the assay is whether the face appears within the outline across a plurality of the sequence of images.