Imaging of biometric information based on three-dimensional shapes

• US 20040223630A1
• Filed: 05/05/2003
• Published: 11/11/2004
• Est. Priority Date: 05/05/2003
• Status: Active Grant

• Current Assignee: Fish & Richardson PC
• Original Assignee: Fish & Richardson PC
• Inventors: Medioni, Gerard, Waupotitsch, Roman, Tsoupko-Sitnikov, Mikhail
• Granted Patent: US 7,242,807 B2
• Time in Patent Office: Days
• US Class Current: 382/118
• CPC Class Codes: G06V 20/653 by matching three-dimensional models, e.g. conformal mapping of Riemann surfaces
• CPC Class Codes: G06V 40/169 Holistic features and representations, i.e. based on the facial image taken as a whole

Abstract

A device for carrying out of biometric identification by comparing face shapes. The shape of a first face is compared with the shape of a second face. A map can be determined between the shapes. The shape, and hence the comparison, is substantially independent of pose and lighting, and also parts of the shape can be removed from the comparison in order to enhance the recognition.

Claims

1. A method comprising:

obtaining a template representing a depth map of a known user'"'"'s face, said depth map including a mask that represents part, but not all, of the users face; and

comparing the template to another depth map to determine a match between the known user'"'"'s face, and the another depth map.

2. A method as in claim 1, wherein said comparing the depth map comprises comparing said another depth map to a plurality of templates to identify whether a person associated with the another depth map is among said plurality of templates.

3. A method as in claim 1, wherein said comparing the depth map comprises comparing said another depth map to a single template to verify an identity of a person associated with said another depth map.

4. A method as in claim 1, wherein said mask represents a part of users face excluding the eyes.

5. A method as in claim 1, wherein said mask represents a part of the user'"'"'s face excluding hair.

6. A method as in claim 1, wherein said obtaining comprises obtaining a three-dimensional map, and removing noise from the map.

7. A method as in claim 6, wherein said noise includes spiking noise formed by hair and other similarly sized objects.

8. A method as in claim 1, further comprising processing the template by determining a tip of the nose within the template, and using said tip of the nose of the template to find other face parts.

9. A method as in claim 8, wherein said other face parts include the eyes, and wherein said map represents the user'"'"'s face without the eyes.

10. A method as in claim 8, wherein said determining a tip of the nose comprises determining an intersection between the mask and a tilted plane.

11. A method as in claim 10, wherein said tilted plane is tilted at 7 degrees.

12. A method as in claim 8, wherein said processing comprises using the nose to determine edges of the user'"'"'s head.

13. A method as in claim 8, wherein said detecting a nose tip comprises detecting a plurality of candidate locations for nose tips, determining likely locations of eyes from said nose tip locations, and analyzing said likely locations to determine if eyes are located in said likely locations.

14. A method of forming a mask of the users face, comprising:

Obtaining information indicative of a depth map of a face;

establishing a specified feature on the face, and determining a location of the feature on the depth map;

verifying that the location actually represents the position of the specified feature; and

forming a depth map of the face that uses the position as a reference point.

15. A method as in claim 14, wherein said specified feature is the nose.

16. A method as in claim 15, wherein said verifying comprises determining positions of the eyes, and determining if the eyes are at a correct point in the face that agrees with the nose.

17. A method as in claim 16, wherein said forming a depth mask comprises removing eye positions from the depth mask.

18. A method as in claim 14, wherein said verifying comprises determining a position a distance between said specified feature on the face and another point on the face, analyzing to determine if the another point on the face is in a proper location relative to the specified feature on the face.

19. A method as in claim 14, wherein said determining comprises determining a plurality of candidate positions for the specified feature on the face, and the verifying comprises determining which of the candidate positions is most likely to be the actual specified feature.

20. A method as in claim 18, wherein said verifying comprises looking for positions of another feature on the face.

21. A method, comprising:

enrolling at least one user by forming a three-dimensional depth mask representing a shape of a part of the users face, and forming a template indicative thereof;

determining another shape of another users face as a depth mask that represents said another shape;

determining known points in both said template and said another shape; and

determining if said another shade corresponds to said template by determining errors there between.

22. A method as in claim 21, wherein said three-dimensional depth mask represents a shape of the users face excluding eye parts.

23. A method as in claim 21, wherein said three-dimensional depth mask represents a shape of the users face excluding hair and excluding eye parts.

24. A method as in claim 21, wherein said determining comprises using ICP.

25. A system, comprising:

a shape determining device, which images a current image to determine information which can be used to form a depth mask of a face within the current image;

a processor, which processes said information to form a depth mask representing at least part of said face, and to compare said depth mask with another depth mask to determine whether said current depth mask and said another depth mask represents the same face.

26. A system as in claim 25, wherein said shape determining device comprises one of a stereo camera a laser or a projected pattern device.

27. A system as in claim 25, wherein said processor produces a depth mask of the user'"'"'s face, which does not include information about the users eyes.

28. A system as in claim 25, wherein said processor compares said depth mask with said another depth mask by finding positions of known parts within each of said depth masks.

29. A system as in claim 28, wherein said known parts include the nose.

30. A method of biometric recognition, comprising:

obtaining a template representing a shape of at least a part of a known face including three-dimensional information about the known face;

obtaining a sample representing a shape of an unknown face including three-dimensional information about the unknown face;

finding reference points in both said template and said sample, and comparing only a part of the shape of the template to only a part of the sample, to determine a comparison between said only a part of the shape template and the only a part of the sample.

31. A method as in claim 30, wherein said reference points comprises tips of the noses in the template and sample.

32. A method as in claim 30, wherein said only a part excludes portions surrounding the eyes.