3D polarimetric face recognition system

Patent metadata

Field	Value
Publication number	US-10915737-B2
Application number	US-201916291066-A
Country	US
Kind code	B2
Filing date	Mar 4, 2019
Priority date	Mar 4, 2019
Publication date	Feb 9, 2021
Grant date	Feb 9, 2021

Abstract

Object/facial recognition systems, methods, architectures, mechanisms and/or apparatus for 3D face or object recognition using a 3D surface reconstructed from polarimetric thermal facial/object imager to form a 2D or 3D image which is rotated to conform with a pose of a reference image to allow comparison thereto.

First claim

The invention claimed is:

1. A computer implemented object recognition method, comprising:
acquiring or calculating long-wave-infrared radiation (LWIR) Stokes images S 0 , S 1 and S 2 using polarimetric imaging of an object of interest;
generating a 3D Stokes image S 0 surface overlay of the object of interest using the acquired Stokes images;
estimating a pose of the 3D Stokes image S 0 surface overlay of the object of interest using rotation (Euler) angles (α i , β i , γ i );
rotating the 3D Stokes image S 0 surface overlay such that the pose of the object of interest provided therein substantially conforms to a pose of a reference object; and
generating a heat map of the rotated 3D Stokes image S 0 surface overlay.

2. The method of claim 1 , wherein the heat map of the rotated 3D Stokes image S 0 surface overlay of the object of interest comprises a 3D heat map of the object of interest and the method further comprises: comparing the 3D heat map of the object of interest to the reference object to generate thereby a similarity score.

3. The method of claim 2 , wherein said step of comparing comprises comparing the generated 3D heat map of the object of interest with a 3D surface of the reference object.

4. The method of claim 2 , wherein said step of comparing is performed by a remote server in response to receiving 3D surface imagery of the object of interest.

5. The method of claim 1 , further comprising:
projecting the heat map of the rotated 3D Stokes image S 0 surface overlay of the object of interest onto a plane perpendicular to the imaging axis to generate a 2D rotated heat map of the object of interest,
wherein said step of comparing comprises comparing the 2D rotated heat map of the object of interest with a 2D surface of the reference object.

6. The method of claim 5 , wherein said step of comparing is performed by a remote server in response to receiving 2D surface imagery of the object of interest.

7. The method of claim 5 ,
wherein the step of comparing further comprises: comparing the 2D heat map of the object of interest with the 2D surface of the reference object.

8. The method of claim 7 , wherein said step of rotating the 3D Stokes image S 0 surface overlay is performed by maximizing a similarity between the pose of the object of interest provided therein and the pose of a reference object.

9. The method of claim 7 , wherein said step of rotating the 3D Stokes image S 0 surface overlay is performed by maximizing a left/right symmetry of the pose of the object of interest.

10. The method of claim 1 , wherein said step of rotating the 3D Stokes image S 0 surface overlay is performed by maximizing a similarity between the pose of the object of interest provided therein and the pose of a reference object.

11. The method of claim 1 , wherein said step of rotating the 3D Stokes image S 0 surface overlay is performed by maximizing a left/right symmetry of the pose of the object of interest.

12. The method of claim 1 , wherein said steps of rotating and comparing are performed for each of a plurality of reference objects within a reference object database, said method further comprising identifying a reference object associated with the highest similarity score.

13. The method of claim 1 , wherein said steps of rotating and comparing are performed for a specific of reference object within a reference object database, said method further comprising identifying a reference object to match said object of interest in response to a similarity score exceeding a threshold level.

14. An apparatus for object recognition, the apparatus comprising a processor configured for:
acquiring or calculating a Stokes images S 0 , S 1 and S 2 using polarimetric sensor information associated with an object of interest;
generating a 3D Stokes image S 0 surface overlay of the object of interest using the acquiring Stokes images;
estimating a pose of the 3D Stokes image S 0 surface overlay of the object of interest using rotation (Euler) angles (α i , β i , γ i );
rotating the 3D Stokes image S 0 surface overlay such that the pose of the object of interest provided therein substantially conforms to a pose of a reference object; and
generating a heat map of the rotated 3D Stokes image S 0 surface overlay.

15. The apparatus of claim 14 , wherein said processor being further configured to compare the generated heat map of the rotated 3D Stokes image S 0 surface overlay of the object of interest to the reference object to generate thereby a similarity score.

16. The apparatus of claim 15 , wherein said comparing comprises comparing the generated heat map of the rotated 3D Stokes image S 0 surface overlay of the object of interest with a 3D surface of the reference object.

17. The apparatus of claim 15 , wherein said processor being further configured to project the heat map of the rotated 3D Stokes image S 0 surface overlay onto a plane perpendicular to the imaging axis to generate a 2D rotated heat map, and said comparing comprises comparing the 2D rotated heat map of the object of interest with a 2D surface of the reference object.

18. The apparatus of claim 15 ,
wherein the step of comparing further comprises comparing the 2D heat map of the rotated Stokes image S 0 surface overlay of the object of interest with the 2D surface of the reference object.

19. The apparatus of claim 15 , wherein said steps of rotating and comparing are performed for each of a plurality of reference objects within a reference object database, said processor being further configured for identifying a reference object associated with the highest similarity score.

20. The apparatus of claim 15 , wherein said steps of rotating and comparing are performed for a specific of reference object within a reference object database, said processor being further configured for identifying a reference object to match said object of interest in response to a similarity score exceeding a threshold level.

21. The apparatus of claim 14 , wherein said rotating the 3D Stokes image S 0 surface overlay is performed by one or both of maximizing a similarity between the pose of the object of interest provided therein and the pose of a reference object, and maximizing a left/right symmetry of the pose of the object of interest.

22. A tangible and non-transient computer readable storage medium storing instructions which, when executed by a computer, adapt the operation of the computer to provide a method for object recognition, comprising:
acquiring or calculating Stokes images S 0 , S 1 and S 2 using polarimetric sensor information associated with an object of interest;
generating a 3D Stokes image S 0 surface overlay of the object of interest using the acquiring Stokes images;
estimating a pose of the 3D Stokes image S 0 surface overlay of the object of interest
rotating the 3D Stokes image S 0 surface overlay such that the pose of the object of interest provided therein substantially conforms to a pose of a reference object; and
generating a 2D or 3D heat map of the rotated 3D Stokes image S 0 surface overlay.

Assignees

• Us Army Res Lab
• Us Army

Inventors

• Hu Shuowen
• Short Nathaniel J
• Yuffa Alex J
• Videen Gorden
• Gurton Kristan P

Classifications

• G06V10/76

  based on eigen-space representations, e.g. from pose or different illumination conditions; Shape manifolds · CPC title

• G06V20/64 Primary

  Three-dimensional [3D] objects · CPC title

• G06V40/172

  Classification, e.g. identification · CPC title

• G01J4/04

  Polarimeters using electric detection means (G01J4/02 takes precedence) · CPC title

• G01J2005/0077

  Imaging · CPC title