Near-infrared method and system for use in face detection

• US 7,027,619 B2
• Filed: 02/15/2002
• Issued: 04/11/2006
• Est. Priority Date: 09/13/2001
• Status: Active Grant

• Current Assignee: Honeywell International Inc.
• Original Assignee: Honeywell International Inc.
• Inventors: Pavlidis, Ioannis, Dowdall, Jonathan B.
• Primary Examiner(s): Mehta, Bhavesh M.
• Assistant Examiner(s): Perungavoor, Sath V.
• Application Number:
  US10/077,672
• Publication Number:
  US 20030053664A1
• Time in Patent Office: 1,516 Days
• Field of Search: 382/103, 382/115, 382/116, 382/117, 382/118, 382/173, 382/190, 382/191, 382/192, 382/193, 382/194, 382/195, 382/205, 382/206, 382/209, 382/216, 382/217, 382/218, 250/330, 250/331, 250/332, 250/333, 250/334, 250/356.1, 250/338.1, 250/339.06, 250/339.11, 250/341.8, 348/77, 348/78, 348/164, 348/169, 348/342, 340/5.53, 340/5.83, 340/545.3
• US Class Current: 382/115
• CPC Class Codes: G06V 10/143 Sensing or illuminating at different wavelengths
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation

Abstract

A method and system for use in detecting a face of a person includes providing image data representative of reflected light of a scene in at least one bandwidth within a reflected infrared radiation range. Regions of the scene in the image data which may correspond to one or both eyebrows of a person and which may correspond to one or both eyes of the person are extracted from the data. The extracted regions which may correspond to one or both eyebrows of the person are combined with the extracted regions that may correspond to one or both eyes of the person resulting in a composite feature image. The composite feature image is searched based on model data representative of an eye region to detect one or both eyes of the person.

Claims

1. A method for use in detecting a face of a person, the method comprising:

providing image data representative of reflected light of a scene in at least a first and second bandwidth within a reflected infrared radiation range;

extracting regions of the scene in the image data representative of reflected light of the scene in the first bandwidth within the reflected infrared radiation range which correspond to one or both eyebrows of a person;

extracting regions of the scene in the image data representative of reflected light of the scene in the second bandwidth within the reflected infrared radiation range which correspond to one or both eyes of the person;

combining the extracted regions of the scene which correspond to one or both eyebrows of the person with the extracted regions of the scene that correspond to one or both eyes of the person resulting in a composite feature image; and

searching the composite feature image based on model data representative of an eye region to detect one or both eyes of the person.

2. The method of claim 1, wherein providing image data representative of reflected light of a scene in at least a first and second bandwidth within a reflected infrared radiation range comprises:

providing a first image output representative of reflected light of a scene in a first bandwidth within a reflected infrared radiation range, wherein regions of the scene in the first image output are extracted which correspond to one or both eyebrows of the person, and further wherein the regions that correspond to one or both eyebrows of the person have a comparatively higher reflectivity in the first bandwidth than the reflectivity of human skin; and

providing a second image output representative of reflected light of a scene in a second bandwidth within a reflected infrared radiation range, wherein regions of the scene in the second image output are extracted which correspond to one or both eyes of the person, and further wherein the regions that correspond to one or both eyes of the person have a comparatively lower reflectivity in the second bandwidth than the reflectivity of human skin.

3. The method of claim 2, wherein a wavelength between the first and second bandwidths is about 1.4 microns.

4. The method of claim 3, wherein the second bandwidth is between at least about 0.9 microns and 1.4 microns and the first bandwidth is between 1.4 microns and at least about 1.7 microns.

5. The method of claim 2, wherein the method further comprises defining a facial search area by isolating skin of a person in the scene relative to background based on the first and second image output.

6. The method of claim 5, wherein a difference in reflectance for human skin in the first bandwidth relative to the second bandwidth is greater than a difference in reflectance for objects in the background other than human skin in the first bandwidth relative to the second bandwidth.

7. The method of claim 5, wherein defining the facial search area by isolating skin of the person in the scene comprises performing a weighted difference calculation of spectral power of the pixels of the first image output with spectral power of corresponding pixels of the second image output resulting in a weighted difference output for the pixels.

8. The method of claim 7, wherein defining the facial search area by isolating skin of the person in the scene further comprises comparing the weighted difference output for the pixels to one or more threshold values to differentiate pixels representative of skin from pixels representative of objects in the background other than human skin.

9. The method of claim 1, wherein the method further comprises defining a facial search area by isolating skin of a person in die scene relative to background.

10. The method of claim 9, wherein regions of the scene in the image data which correspond to one or both eyebrows of the person or which correspond to one or both eyes of the person are extracted from within the defined search area.

11. The method of claim 9, wherein the method further comprises detecting one of orientation or extent of the face of the person based on the detected one or both eyes of the person and the isolated skin of the person.

12. The method of claim 1, wherein the method further comprises detecting the orientation of the face of the person based at least on whether one or both eyes are detected.

13. The method of claim 1, wherein searching the composite feature image based on model data representative of an eye region to detect one or both eyes of the person comprises:

searching the composite feature image based on model data representative of an eye region resulting in one or more eye containing regions; and

estimating the center of one or both eyes within the one or more eye containing regions.

14. The method of claim 13, wherein searching the composite feature image based on model data representative of an eye region resulting in one or more eye containing regions comprises applying a template representative of a relationship between eyes and eyebrows of a plurality of persons to the composite feature image to locate the one or more eye containing regions.

15. The method of claim 13, wherein estimating the center of one or both eyes within the one or more eye containing regions comprises determining whether the one or more eye containing regions are a single blob that spans a major portion of the face of the person, two blobs of about equivalent size, or a single blob that is relatively smaller than the size of the single blob that spans a major portion of the face of the person but is larger than other eye containing regions located.

16. The method of claim 1, wherein the method further comprises monitoring luminance level in the scene and dynamically adjusting the luminance level in response thereto.

17. A system for use in detecting a face of a person, the system comprising:

at least one imaging device operable to provide image data representative of reflected light of a scene in at least a first and second bandwidth within a reflected infrared radiation range; and

processing apparatus operable to;

extract regions of the scene in the image data representative of reflected light of the scene in the first bandwidth within the reflected infrared radiation range which correspond to one or both eyebrows of a person;

extract regions of the scene in the image data representative of reflected light of the scene in the second bandwidth within the reflected infrared radiation range which correspond to one or both eyes of the person;

combine the extracted regions of the scene which correspond to one or both eyebrows of the person with the extracted regions of the scene that correspond to one or both eyes of the person resulting in a composite feature image; and

search the composite feature image based on model data representative of an eye region to detect one or both eyes of the person.

18. The system of claim 17, wherein the at least one imaging device comprises:

a first imaging device sensitive to light of the first bandwidth within a reflected infrared radiation range to provide a first image output representative of the scene, wherein regions of the scene in the first image output are extracted which correspond to one or both eyebrows of the person, and further wherein the regions that correspond to one or both eyebrows of the person have a comparatively higher reflectivity in the first bandwidth than the reflectivity of human skin; and

a second imaging device sensitive to light of the second bandwidth within the reflected infrared radiation range to provide a second image output representative of the scene, wherein regions of the scene in the second image output are extracted which correspond to one or both eyes of the person, and further wherein the regions that correspond to one or both eyes of the person have a comparatively lower reflectivity in the first bandwidth than the reflectivity of human skin.

19. The system of claim 18, wherein a wavelength between the first and second bandwidths is about 1.4 microns.

20. The system of claim 18, wherein the second bandwidth is between at least about 0.9 microns and 1.4 microns and the first bandwidth is between 1.4 microns and at least about 1.7 microns.

21. The system of claim 18, wherein the processing apparatus is further operable to define a facial search area by isolating skin of a person in the scene relative to background based on the first and second image output.

22. The system of claim 21, wherein a difference in reflectance for human skin in the first bandwidth relative to the second bandwidth is greater than a difference in reflectance for objects in the background other than human skin in the first bandwidth relative to the second bandwidth.

23. The system of claim 21, wherein the processing apparatus is further operable to perform a weighted difference calculation of spectral power of the pixels of the first image output with spectral power of corresponding pixels of the second image output resulting in a weighted difference output for the pixels.

24. The system of claim 23, wherein the processing apparatus is further operable to compare the weighted difference output for the pixels to one or more threshold values to differentiate pixels representative of skin from pixels representative of objects in the background other than human skin.

25. The system of claim 17, wherein the processing apparatus is further operable to define a facial search area by isolating skin of a person in the scene relative to background.

26. The system of claim 25, wherein regions of the scene in die image data which correspond to one or both eyebrows of the person or which correspond to one or both eyes of the person are extracted from within the defined facial search area.

27. The system of claim 25, wherein the processing apparatus is further operable to detect one of orientation or extent of the face of the person based on the detected one or both eyes of the person and the isolated skin of the person.

28. The system of claim 17, wherein the processing apparatus is further operable to detect the orientation of the face of the person based at least on whether one or both eyes are detected.

29. The system of claim 17, wherein the processing apparatus is further operable to:

search the composite feature image based on model data representative of an eye region resulting in one or more eye containing regions; and

estimate the center of one or both eyes within the one or more eye containing regions.

30. The system of claim 29, wherein the processing apparatus is further operable to apply a template representative of a relationship between eyes and eyebrows of a plurality of persons to the composite feature image to locate the one or more eye containing regions.

31. The system of claim 29, wherein the processing apparatus is further operable to determine whether the one or more eye containing regions are a single blob that spans a major portion of the face of the person, two blobs of about equivalent size, or a single blob that is relatively smaller than the size of the single blob that spans a major portion of the face of the person but is larger than other eye containing regions located.

32. The system of claim 17, wherein the system further comprises a detection device operable to monitor luminance level in the scene and control means to dynamically adjust the luminance level provided by a light source in response thereto.

33. A method for use in detecting a face of a person, the method comprising:

providing a first image output representative of reflected light of a scene in a first bandwidth within a reflected infrared radiation range;

providing a second image output representative of reflected light of a scene in a second bandwidth within a reflected infrared radiation range;

defining at least a facial search area by isolating skin of a person in the scene relative to background based on the first and second image output, wherein a difference in reflectance for human skin in the first bandwidth relative to the second bandwidth is greater than a difference in reflectance for objects in the background other than human skin in the first bandwidth relative to the second bandwidth;

extracting regions in the defined facial search area which correspond to one or both eyebrows of the person, wherein the regions that correspond to one or both eyebrows of the person have a comparatively higher reflectivity in the first bandwidth than the reflectivity of human skin; and

extracting regions in the defined facial search area which correspond to one or both eyes of the person, wherein the regions that correspond to one or both eyes of the person have a comparatively lower reflectivity in the second bandwidth than the reflectivity of human skin;

combining the extracted regions of the scene which correspond to one or both eyebrows of the person with the extracted regions of the scene that correspond to one or both eyes of the person resulting in a composite feature image;

searching the composite feature image based on model data representative of an eye region to locate one or more eye containing regions; and

estimating the center of one or both eyes within the one or more eye containing regions.

34. The method of claim 33, wherein a wavelength between the first and second bandwidths is about 1.4 microns.

35. The method of claim 33, wherein the second bandwidth is between at least about 0.9 microns and 1.4 microns and the first bandwidth is between 1.4 microns and at least about 1.7 microns.

36. The method of claim 33, wherein defining the facial search area by isolating skin of the person in the scene comprises performing a weighted difference calculation of spectral power of the pixels of the first image output with spectral power of corresponding pixels of the second image output resulting in a weighted difference output for the pixels.

37. The method of claim 36, wherein defining the facial search area by isolating skin of the person in the scene further comprises comparing the weighted difference output for the pixels to one or more threshold values to differentiate pixels representative of skin from pixels representative of objects in the background other than human skin.

38. The method of claim 33, wherein the method further comprises detecting one of orientation or extent of the face of the person based on the detected one or both eyes of the person and the isolated skin of the person.

39. The method of claim 33, wherein the method further comprises detecting the orientation of the face of the person based at least on whether one or both eyes are detected.

40. The method of claim 33, wherein searching the composite feature image based on model data representative of an eye region resulting in one or more eye containing regions comprises applying a template representative of a relationship between eyes and eyebrows of a plurality of persons to the composite feature image to locate the one or more eye containing regions.

41. The method of claim 33, wherein estimating the center of one or both eyes within the one or more eye containing regions comprises determining whether the one or more eye containing regions are a single blob that spans a major portion of the face of the person, two blobs of about equivalent size, or a single blob that is relatively smaller than the size of the single blob that spans a major portion of the face of the person but is larger than other eye containing regions located.

42. The method of claim 33, wherein the method further comprises applying at least data representative of the estimated center of the one or both eyes to a face recognition system.