Facial spoofing detection in image based biometrics

• US 9,690,998 B2
• Filed: 12/19/2014
• Issued: 06/27/2017
• 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: Negi, Ansuya, Graumann, David L., Ghosh, Rahuldeva, Narjala, Ranjit S.
• Primary Examiner(s): Chu, Randolph I
• Application Number:
  US14/779,842
• Publication Number:
  US 20160328621A1
• Time in Patent Office: 921 Days
• Field of Search: None
• CPC Class Codes: G06V 10/44 Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
• CPC Class Codes: G06V 40/16 Human faces, e.g. facial parts, sketches or expressions
• CPC Class Codes: G06V 40/40 Spoof detection, e.g. liveness detection

Abstract

Systems and techniques for facial spoofing detection in image based biometrics are described herein. A marker may be created for a representation of a face in a first plurality of images of a sequence of images. The marker corresponds to a facial feature of the face. An environmental feature of an environment of the face may be identified across a second plurality of images of the sequence of images. A correlation between the marker and the environmental feature in the sequence of images may be quantified to produce a synchronicity metric. A spoofing attempt may be indicated in response to the synchronicity metric meeting a threshold.

Claims

1. A system for facial spoofing detection in image based biometrics, the method comprising:

a biometric feature detector to create a marker for a representation of a face in a first plurality of images of a sequence of images, the marker corresponding to a facial feature;

an environmental feature detector to identify an environmental feature of an environment of the face across a second plurality of images of the sequence of images, wherein the environmental feature includes edges detected in the second plurality of images, wherein the edges are moving edges, and wherein the edges define a convex hull;

a synchronicity detector to quantify a correlation between the marker and the environmental feature in the sequence of images to produce a synchronicity metric, wherein the convex hull does not include the marker resulting in the synchronicity metric being reduced; and

a spoofing indication controller to indicate a spoofing attempt in response to the synchronicity metric meeting a threshold.

2. The system of claim 1, wherein the marker is a line between eyes in the representation of the face.

3. The system of claim 1, wherein the first plurality of images does not include a member of the second plurality of images, wherein the member precedes in time the first plurality of images and includes a determined resting spot for the moving edges, the moving edges outlining an area, and wherein the marker is within the area for an image in which the moving edges are not detected following the member, the marker is determined to be within the convex hull for the synchronicity metric.

4. The system of claim 1, wherein the environmental feature detector is to compare the edges to a catalog of device edge configurations to determine whether the edges represent a known device.

5. The system of claim 4, wherein the edges correspond to a known device, and wherein a subset of edges correspond to a display area of the known device, and wherein the display area does not include the marker resulting in the synchronicity metric being reduced.

16. The system of claim 1, comprising an image capture component to:

obtain a sequence of raw images from a camera; and

modify the sequence of raw images to produce the sequence of images, the modifying including changing at least one of gain, exposure, or gamma to enhance edge detection.

17. The system of claim 16, wherein to modify the sequence of raw images includes the image capture component to apply a noise filter to the sequence of raw images to produce the sequence of images.

18. The system of claim 1, wherein the sequence of images is at least thirty images.

6. A method for facial spoofing detection in image based biometrics, the method comprising:

creating, using a first group of circuits, a marker for a representation of a face in a first plurality of images of a sequence of images, the marker corresponding to a facial feature;

identifying an environmental feature of an environment of the face across a second plurality of images of the sequence of images using a second group of circuits, wherein the environmental feature includes edges detected in the second plurality of images, wherein the edges are moving edges, and wherein the edges define a convex hull;

quantifying, using a third group of circuits, a correlation between the marker and the environmental feature in the sequence of images to produce a synchronicity metric, wherein the convex hull does not include the marker resulting in the synchronicity metric being reduced; and

indicating, using a fourth group of circuits, a spoofing attempt in response to the synchronicity metric meeting a threshold.

7. The method of claim 6, wherein the marker is a line between eyes in the representation of the face.

8. The method of claim 6, wherein the first plurality of images does not include a member of the second plurality of images, wherein the member precedes in time the first plurality of images and includes a determined resting spot for the moving edges, the moving edges outlining an area, and wherein the marker is within area for an image in which the moving edges are not detected following the member, the marker is determined to be within the convex hull for the synchronicity metric.

9. The method of claim 6, wherein the edges are compared to a catalog of device edge configurations to determine whether the edges represent a known device.

10. The method of claim 9, wherein the edges correspond to a known device, and wherein a subset of edges correspond to a display area of the known device, and wherein the display area does not include the marker resulting in the synchronicity metric being reduced.

19. The method of claim 6, comprising:

obtaining a sequence of raw images from a camera; and

modifying the sequence of raw images to produce the sequence of images, the modifying including changing at least one of gain, exposure, or gamma to enhance edge detection.

20. The method of claim 19, wherein modifying the sequence of raw images includes applying a noise filter to the sequence of raw images to produce the sequence of images.

21. The method of claim 6, wherein the sequence of images is at least thirty images.

11. At least one machine readable medium that is not a transitory propagating signal, the at least one machine readable medium including instructions that, when executed by a machine, cause the machine to perform operations:

creating, using a first group of circuits, a marker for a representation of a face in a first plurality of images of a sequence of images, the marker corresponding to a facial feature;

identifying an environmental feature of an environment of the face across a second plurality of images of the sequence of images using a second group of circuits, wherein the environmental feature includes edges detected in the second plurality of images, wherein the edges are moving edges, and wherein the edges define a convex hull;

quantifying, using a third group of circuits, a correlation between the marker and the environmental feature in the sequence of images to produce a synchronicity metric, wherein the convex hull does not include the marker resulting in the synchronicity metric being reduced; and

indicating, using a fourth group of circuits, a spoofing attempt in response to the synchronicity metric meeting a threshold.

12. The at least one machine readable medium of claim 11, wherein the marker is a line between eyes in the representation of the face.

13. The at least one machine readable medium of claim 11, wherein the first plurality of images does not include a member of the second plurality of images, wherein the member precedes in time the first plurality of images and includes a determined resting spot for the moving edges, the moving edges outlining an area, and wherein the marker is within area for an image in which the moving edges are not detected following the member, the marker is determined to be within the convex hull for the synchronicity metric.

14. The at least one machine readable medium of claim 11, wherein the edges are compared to a catalog of device edge configurations to determine whether the edges represent a known device.

15. The at least one machine readable medium of claim 14, wherein the edges correspond to a known device, and wherein a subset of edges correspond to a display area of the known device, and wherein the display area does not include the marker resulting in the synchronicity metric being reduced.

22. The at least one machine readable medium of claim 11, wherein the operations comprise:

obtaining a sequence of raw images from a camera; and

modifying the sequence of raw images to produce the sequence of images, the modifying including changing at least one of gain, exposure, or gamma to enhance edge detection.

23. The at least one machine readable medium of claim 22, wherein modifying the sequence of raw images includes the image capture component to apply a noise filter to the sequence of raw images to produce the sequence of images.

24. The at least one machine readable medium of claim 11, wherein the sequence of images is at least thirty images.