Face recognition apparatus, method, system and computer readable medium thereof

• US 6,108,437 A
• Filed: 02/20/1998
• Issued: 08/22/2000
• Est. Priority Date: 11/14/1997
• Status: Expired due to Term

• Current Assignee: Seiko Epson Corporation (Seiko Group)
• Original Assignee: Seiko Epson Corporation (Seiko Group)
• Inventors: Lin, Shang-Hung
• Primary Examiner(s): Chang, Jon
• Assistant Examiner(s): Bali, Vikkram
• Application Number:
  US09/026,970
• Time in Patent Office: 914 Days
• Field of Search: 382/117, 382/118, 382/157, 382/291, 382/116, 382/155, 382/156, 382/158, 382/159, 382/190, 382/209, 382/199, 382/217, 382/218, 340/825.34, 348/77, 348/78, 380/23
• US Class Current: 382/118
• CPC Class Codes: G06V 40/16 Human faces, e.g. facial parts, sketches or expressions

Abstract

A face recognition system is provided comprising an input process or circuit, such as a video camera for generating an image of a person. A face detector process or circuit determines if a face is present in a image. A face position registration process or circuit determines a position of the face in the image if the face detector process or circuit determines that the face is present. A feature extractor process or circuit is provided for extracting at least two facial features from the face. A voting process or circuit compares the extractor facial features with a database of extracted facial features to identify the face.

Claims

1. A face recognition system comprising:

an input unit for inputting an image of a subject;

a face detector for determining if a face is present in the image from said input circuit;

a face position registration circuit for determining a position of the face in the image if said face detector determines that the face is present, said face position registration circuit comprising an eye template circuit that applies an eye template to a first side of the face and a mirror image of the eye template to a second side of the face;

a feature extractor for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration circuit; and

a voting circuit for comparing the at least two extracted facial features with a database of candidates corresponding facial features to identify the face.

2. A face recognition system according to claim 1, wherein said face detector comprises:

a lumen normalization circuit for lumen normalizing the output by said input circuit;

an edge extractor for extracting edge information from an output of said lumen normalization circuit;

a down sampling circuit for down sampling at a predetermined ratio an output signal from said edge extraction circuit;

a facial template circuit for applying a plurality of facial templates to an output of said down sampling circuit; and

a facial template matching circuit for determining which one of the plurality of facial templates corresponds to the output of said down sampling circuit.

3. A face recognition system according to claim 2,

wherein the image output by said input device comprises M×

N pixels, wherein M and N are positive integers,

wherein said down sampling circuit comprises;

a low pass filter to filter the M×

N pixels;

a down sampler for sampling every n^th pixel from said low pass filter to output M/n×

N/n pixels, wherein n is a positive integer.

4. A face recognition system according to claim 1, wherein said face position registration circuit comprises:

a lumen normalization circuit for lumen normalizing an output by said face detector circuit;

an edge extractor for extracting edge information from an output of said lumen normalization circuit;

a down sampling circuit for down sampling at a first predetermined ratio an output signal from said edge extraction circuit;

an eye template circuit for applying a plurality of eye templates to an output of said down sampling circuit;

an eye template matching circuit for determining which one of the plurality of eye templates corresponds to the output of said down sampling circuit; and

face alignment circuit for aligning the face of the image output by said input unit in accordance with said eye matching circuit.

6. A face recognition system according to claim 4, wherein said eye template circuit applies an eye template comprising a right eye and a left eye.

24. A face recognition system according to claim 1, wherein said input unit comprises a video camera.

25. A face recognition system according to claim 1, wherein said input unit comprises a digital camera.

26. A face recognition system according to claim 1, wherein said input unit comprises a memory comprising an image.

27. A face recognition system according to claim 1, wherein the first facial feature comprises a single eye.

28. A face recognition system according to claim 1, wherein the second facial feature comprises both eyes and nose.

29. A face recognition system according to claim 1, wherein said face detector comprises a neural network.

30. A face recognition system according to claim 1, wherein said face position registration circuit comprises a neural network.

31. A face recognition system according to claim 1, wherein said feature extractor comprises a neural network.

5. A face recognition system comprising:

an input unit for inputting an image of a subject;

a face detector for determining if a face is present in the image from said input circuit;

a face position registration circuit for determining a position of the face in the image if said face detector determines that the face is present, said face position registration circuit comprising;

a lumen normalization circuit for lumen normalizing an output by said face detector circuit;

an edge extractor for extracting edge information from an output of said lumen normalization circuit;

a down sampling circuit for down sampling at a first predetermined ratio an output signal from said edge extraction circuit;

an eye template circuit for applying a plurality of eye templates to an output of said down sampling circuit;

an eye template matching circuit for determining which one of the plurality of eye templates corresponds to the output of said down sampling circuit; and

a face alignment circuit for aligning the face of the image output by said input unit in accordance with said eye matching circuit;

a feature extractor for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration circuit; and

a voting circuit for comparing the at least two extracted facial features with a database of candidates corresponding facial features to identify the face;

wherein said eye template circuit applies an eye template to a first side of the face and a mirror image of the eye template to a second side of the face.

7. A face recognition system comprising:

an input unit for inputting an image of a subject;

a face detector for determining if a face is present in the image from said input circuit;

a face position registration circuit for determining a position of the face in the image if said face detector determines that the face is present, said face position registration circuit comprising;

a lumen normalization circuit for lumen normalizing an output by said face detector circuit;

an edge extractor for extracting edge information from an output of said lumen normalization circuit;

a down sampling circuit for down sampling at a first predetermined ratio an output signal from said edge extraction circuit;

an eye template circuit for applying a plurality of eye templates to an output of said down sampling circuit;

an eye template matching circuit for determining which one of the plurality of eye templates corresponds to the output of said down sampling circuit; and

face alignment circuit for aligning the face of the image output by said input unit in accordance with said eye matching circuit;

a feature extractor for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration circuit; and

a voting circuit for comparing the at least two extracted facial features with a database of candidates corresponding facial features to identify the face;

wherein if said eye matching circuit does not determine that any one of the plurality of eye templates corresponds to the output of said down sampling circuit, said down sampling circuit down samples at a second predetermined ratio the output signal from said edge extraction circuit.

8. A face recognition system comprising:

an input unit for inputting an image of a subject;

a face detector for determining if a face is present in the image from said input circuit;

a face position registration circuit for determining a position of the face in the image if said face detector determines that the face is present;

a feature extractor for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration circuit; and

a voting circuit for comparing the at least two extracted facial features with a database of candidates corresponding facial features to identify the face;

wherein said feature extractor comprises;

a cropping circuit for cropping at least the first facial feature and the second facial feature output by said face position registration circuit;

a lumen normalization circuit for lumen normalizing the first facial feature and the second facial feature output by said cropping circuit;

an edge extractor for extracting edge information from the first facial feature and the second facial feature output by said lumen normalization circuit;

a down sampling circuit for down sampling at a first predetermined ratio the first facial feature and at a second predetermined ratio the second facial feature output signal by said edge extraction circuit; and

a feature vector calculating circuit for calculating a first feature vector of the first facial feature from said down sampling circuit and a second feature vector of the second facial feature output from said down sampling circuit.

9. A face recognition system according to claim 8, wherein the first predetermined ratio is greater than the second predetermined ratio.

10. A face recognition system according to claim 8, wherein the first facial feature comprises a single eye.

11. A face recognition system according to claim 8, wherein the second facial feature comprises both eyes and nose.

12. A face recognition system according to claim 8,

wherein said cropping circuit crops a third facial output by said face position registration circuit;

wherein said lumen normalization circuit lumen normalizes the third facial feature output by said cropping circuit detector circuit;

wherein said edge extractor extracts edge information from the third facial feature output by said lumen normalization circuit;

wherein said down sampling circuit down samples at a third predetermined ratio the third facial feature output by said extraction circuit; and

wherein said feature vector calculating circuit calculates a third feature vector of the third facial feature from said down sampling circuit.

13. A face recognition system according to claim 12, wherein the third facial feature comprises both eyes, nose and mouth.

14. A face recognition system according to claim 12, wherein the first predetermined ratio is greater than the second predetermined ratio, and wherein the second ratio is greater than the third ratio.

15. A face recognition system according to claim 8, wherein said voting circuit comprises:

a first recognizer circuit for comparing the first feature vector with corresponding first feature vectors of the database of candidates;

a second recognizer circuit for comparing the second feature vector with corresponding second feature vectors of the database of candidates; and

a selection circuit of selecting one of the candidates in the database in accordance with one of said first recognizer circuit and said second recognizer circuit.

16. A face recognition system comprising:

an input unit for inputting an image of a subject;

a face detector for determining if a face is present in the image from said input circuit;

a face position registration circuit for determining a position of the face in the image if said face detector determines that the face is present;

a feature extractor for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration circuit; and

a voting circuit for comparing the at least two extracted facial features with a database of candidates corresponding facial features to identify the face;

wherein said feature extractor comprises;

a cropping circuit for cropping at least the first facial feature and the second facial feature output by said face position registration circuit;

a down sampling circuit for down sampling at a first predetermined ratio the first facial feature and a second predetermined ratio the second facial feature output signal by said cropping circuit; and

a feature vector calculating circuit for calculating a first feature vector of the first facial feature from said down sampling circuit and a second feature vector of the second facial feature output from said down sampling circuit.

17. A face recognition system according to claim 16, wherein the first predetermined ratio is greater than the second predetermined ratio.

18. A face recognition system according to claim 16,

wherein said cropping circuit crops a third facial output by said face position registration circuit;

wherein said down sampling circuit down samples at a third predetermined ratio the third facial feature output by said cropping circuit; and

wherein said feature vector calculating circuit calculates a third feature vector of the third facial feature from said down sampling circuit.

19. A face recognition system according to claim 18, wherein the first predetermined ratio is greater than the second predetermined ratio, and wherein the second ratio is greater than the third ratio.

20. A face recognition system according to claim 18, wherein the third facial feature comprises both eyes, nose and mouth.

21. A face recognition system according to claim 16, wherein said voting circuit comprises:

a first recognizer circuit for comparing the first feature vector with corresponding first feature vectors of a database of candidates;

a second recognizer circuit for comparing the second feature vector with corresponding second feature vectors of the database of candidates; and

a selection circuit of selecting one of the candidates in the database in accordance with one of said first recognizer circuit and said second recognizer circuit.

22. A face recognition system according to claim 16, wherein the first facial feature comprises a single eye.

23. A face recognition system according to claim 16, wherein the second facial feature comprises both eyes and nose.

32. A face recognition system comprising:

an input unit for inputting an image of a subject;

a face position registration circuit for determining a position of a face in the image from said input circuit, said face position registration circuit comprising an eye template circuit for applying an eye template to a first side of the face and a mirror image of the eye template to a second side of the face;

a feature extractor for extracting at least first and second facial features from the face; and

a voting circuit for comparing the first and second facial features extracted by said feature extractor circuit with a database of extracted facial features to identify the face.

33. A face recognition system according to claim 32, wherein said input unit comprises a video camera.

34. A face recognition system according to claim 32, wherein said input unit comprises a digital camera.

35. A face recognition system according to claim 32, wherein said input unit comprises a memory comprising an image.

36. A face recognition system according to claim 32, wherein the first facial feature comprises a single eye.

37. A face recognition system according to claim 32, wherein the second facial feature comprises both eyes and nose.

38. A face recognition system according to claim 32, wherein said face position registration circuit comprises a neural network.

39. A face recognition system according to claim 32, wherein said feature extractor comprises a neural network.

40. A method of face recognition comprising the steps of:

(a) inputting an image of a subject;

(b) determining if a face is present in the image output of step (a);

(c) determining a position of the face in the image if step (b) determines that the face is present including applying an eye template to a first side of the face and a mirror image of the eye template to a second side of the face;

(d) extracting at least first and second facial features from the face output by step (c); and

(e) comparing the first and second facial features output by step (d) to identify the face with a database of facial features.

41. A method according to claim 40, wherein step (b) comprises the steps of:

(b1) lumen normalizing the output of step (a);

(b2) extracting edge information from an output step (b1);

(b3) down sampling at a first predetermined ratio an output of step (b2);

(b4) applying a plurality of facial templates to an output of step (b3); and

(b5) determining which one of the plurality of facial templates corresponds to the output of step (b4).

42. A method according to claim 41,

wherein the output of step (b2) comprises M×

N pixels, wherein M and N are positive integers,

wherein step (b2) comprises the steps of;

(b2I) low pass filtering the M×

N pixels;

(b2II) sampling every n^th pixel from (b2I) to output M/n×

M/n pixels, wherein n is a position integer.

43. A method according to claim 40, wherein step (c) comprises the steps of:

(c1) lumen normalizing an output of step (b);

(c2) extracting edge information from an output of step (c1);

(c3) down sampling at a first predetermined ratio an output from step (c2);

(c4) applying plurality of eye templates to an output step (c3);

(c5) determining which one of the plurality of eye templates corresponds to the output of step (c3); and

(c6) aligning the face of the image inputted in step (a) in accordance with step (c5).

45. A method according to claim 43, wherein step (c4) applies an eye template comprising a right eye and a left eye.

56. A method according to claim 40, wherein step (e) comprises the steps of:

(e1) comparing the first feature extracted in step (d) with corresponding first features of a database of candidates;

(e2) comparing the second feature extracted in step (d) with corresponding second features of the database of candidates; and

(e3) selecting one of the candidates in the database in accordance with one step (e1) and step (e2).

57. A method according to claim 40, wherein step (b) comprises a neural network.

58. A method according to claim 40, wherein step (c) comprises a neural network.

59. A method according to claim 40, wherein step (d) comprises a neural network.

44. A method of face recognition comprising the steps of:

(a) inputting an image of a subject;

(b) determining if a face is present in the image output of step (a);

(c) determining a position of the face in the image if step (b) determines that the face is -present, wherein step (c) comprises the steps of;

(c1) lumen normalizing an output of step (b);

(c2) extracting edge information from an output of step (c1);

(c3) down sampling at a first predetermined ratio an output from step (c2);

(c4) applying a plurality of eye templates to an output of step (c3);

(c5) determining which one of the plurality of eye templates corresponds to the output of step (c3); and

(c6) aligning the face of the image inputted in step (a) in accordance with step (c5);

(d) extracting at least first and second facial features from the face output by step (c); and

(e) comparing the first and second facial features output by step (d) to identify the face with a database of facial features;

wherein step (c4) applies an eye template to a first side of the face and a mirror image of the eye template to a second side of the face.

46. A method of face recognition comprising the steps of:

(a) inputting an image of a subject;

(b) determining if a face is present in the image output of step (a);

(c) determining a position of the face in the image if step (b) determines that the face is present, wherein step (c) comprises the steps of;

(c1) lumen normalizing an output of step (b);

(c2) extracting edge information from an output of step (c1);

(c3) down sampling at a first predetermined ratio an output from step (c2);

(c4) applying a plurality of eye templates to an output of step (c3);

(c5) determining which one of the plurality of eye templates corresponds to the output of step (c3); and

(c6) aligning the face of the image inputted in step (a) in accordance with step (c5);

(d) extracting at least first and second facial features from the face output by step (c); and

(e) comparing the first and second facial features output by step (d) to identify the face with a database of facial features;

wherein if step (c5) does not determine that any one of the plurality of eye templates corresponds to the output of step (c3), step (c3) down samples at a second predetermined ratio the output signal from step (c2).

47. A method of face recognition comprising the steps of:

(a) inputting an image of a subject;

(b) determining if a face is present in the image output of step (a);

(c) determining a position of the face in the image if step (b) determines that the face is present;

(d) extracting at least first and second facial features from the face output by step (c); and

(e) comparing the first and second facial features output by step (d) to identify the face with a database of facial features;

wherein step (d) comprises the steps of;

(d1) cropping at least a first facial feature and a second facial feature output by step (c);

(d2) lumen normalizing the first facial feature and the second facial feature output by step (d1);

(d3) extracting edge information from the first facial feature and the second facial feature output by step (d2);

(d4) down sampling at a first predetermined ratio the first facial feature and at a second predetermined ratio the second facial feature output signal by step (d3); and

(d5) calculating a first feature vector of the first facial feature from step (d4) and a second feature vector of the second facial feature output from step (d4).

48. A method according to claim 47, wherein the first predetermined ratio is greater than the second predetermined ratio.

49. A method according to claim 47, further comprising the steps of

(d6) cropping a third facial output by output by step (c);

(d7) lumen normalizing the third facial feature output by step (d6);

(d8) extracting edge information from the third facial feature output step (d7);

(d9) down sampling at a third predetermined ratio the third facial feature output by step (d8); and

(d10)calculating a third feature vector of the third facial feature from (d9).

50. A method according to claim 49, wherein the first predetermined ratio is greater than the second predetermined ratio, and wherein the second ratio is greater than the third ratio.

55. A method according to claim 47, wherein step (e) comprises the steps of:

(e1) comparing the first feature vector calculated in step (5) with corresponding first feature vectors of a database of candidates;

(e2) comparing the second feature vector calculated in step (5) with corresponding second features of the database of candidates; and

(e3) selecting one of the candidates in the database in accordance with one step (e1) and step (e2).

51. A method of face recognition comprising the steps of:

(a) inputting an image of a subject;

(b) determining if a face is present in the image output of step (a);

(c) determining a position of the face in the image if step (b) determines that the face is present;

(d) extracting at least first and second facial features from the face output by step (c); and

(e) comparing the first and second facial features output by step (d) to identify the face with a database of facial features;

wherein step (d) comprises the steps of;

(d1) cropping at least a first facial feature and a second facial feature output by step (c);

(d2) down sampling at a first predetermined ratio the first facial feature and at a second predetermined ratio the second facial feature output signal by step (d1); and

(d3) calculating a first feature vector of the first facial feature from step (d2) and a second feature vector of the second facial feature output from step (d2).

52. A method according to claim 51, wherein the first predetermined ratio is greater than the second predetermined ratio.

53. A method according to claim 51, further comprising the steps of:

(d4) cropping a third facial output by step (c);

(d5) down sampling at a third predetermined ratio the third facial feature output by step (d4); and

(d6) calculating a third feature vector of the third facial feature from step (d5).

54. A method according to claim 53, wherein the first predetermined ratio is greater than the second predetermined ratio, and wherein the second ratio is greater than the third ratio.

60. A method of face recognition comprising the steps of:

(a) generating an image of a subject;

(b) determining a position of a face in the image from step (a) including applying an eye template to a first side of the face and a mirror image of the eye template to a second side of the face;

(c) extracting at least two facial features from the face determined in step (b); and

(d) comparing the extracted facial features of step (c) with a database of extracted facial features to identify the face.

61. A method according to claim 60, wherein step (b) comprises a neural network.

62. A method according to claim 60, wherein step (c) comprises a neural network.

63. A face recognition system comprising:

input means for inputting an image of a subject;

face detector means for determining if a face is present in the image output by said input means;

face position registration means for determining a position of the face in the image if said face detector means determines that the face is present, said face position registration means comprising eye template means for applying an eye template to a first side of the face and a mirror image of the eye template to a second side of the face;

feature extractor means for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration means; and

voting means for comparing the at least two extracted facial features with a database of corresponding facial features to identify the face.

64. A face recognition system according to claim 63, wherein said face detector means comprises:

Lumen normalization means for lumen normalizing the output by said input means;

edge extraction means for extracting edge information from an output of said lumen normalization means;

down sampling means for down sampling at a predetermined ratio an output signal from said edge extraction means;

facial template means for applying plurality of facial templates to an output of said down sampling means; and

facial template matching means for determining which one of the plurality of facial templates corresponds to the output of said down sampling means.

65. A face recognition system according to claim 64,

wherein the image output by said input means comprises M×

N pixels, wherein M and N are positive integers,

wherein said down sampling means comprises;

low pass filtering means to filter the M×

N pixels;

sampler means for sampling every nth pixel from said low pass filter to output M/n×

N/n pixels, wherein n is a positive integer.

66. A face recognition system according to claim 63, wherein said face position registration means comprises:

Lumen normalization means for lumen normalizing an output by said face detector means;

edge extraction means for extracting edge information from an output of said lumen normalization means;

down sampling means for down sampling at a first predetermined ratio an output signal from said edge extraction means;

eye template means for applying a plurality of eye templates to an output of said down sampling means;

eye template matching means for determining which one of the plurality of eye templates corresponds to the output of said down sampling means; and

face alignment means for aligning the face of the image output by said input means in accordance with said eye matching means.

68. A face recognition system according to claim 66, wherein said eye template means applies an eye template comprising a right eye and a left eye.

80. A face recognition system according to claim 63, wherein said feature extractor means comprises a neural network.

67. A face recognition system comprising:

input means for inputting an image of a subject;

face detector means for determining if a face is present in the image output by said input means;

face position registration means for determining a position of the face in the image if said face detector means determines that the face is present, said face position registration means comprising;

lumen normalization means for lumen normalizing an output by said face detector means;

edge extraction means for extracting edge information from an output of said lumen normalization means;

down sampling means for down sampling at a first predetermined ratio an output signal from said edge extraction means;

eye template means for applying a plurality of eye templates to an output of said down sampling means;

eye template matching means for determining which one of the plurality of eye templates corresponds to the output of said down sampling means; and

face alignment means for aligning the face of the image output by said input means in accordance with said eye matching means;

feature extractor means for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration means; and

voting means for comparing the at least two extracted facial features with a database of corresponding facial features to identify the face;

wherein said eye template means applies an eye template to a first side of the face and a mirror image of the eye template to a second side of the face.

69. A face recognition system comprising:

input means for inputting an image of a subject;

face detector means for determining if a face is present in the image output by said input means;

face position registration means for determining a position of the face in the image if said face detector means determines that the face is present, said face position registration means comprising;

lumen normalization means for lumen normalizing an output by said face detector means;

edge extraction means for extracting edge information from an output of said lumen normalization means;

down sampling means for down sampling at a first predetermined ratio an output signal from said edge extraction means;

eye template means for applying a plurality of eye templates to an output of said down sampling means;

eye template matching means for determining which one of the plurality of eye templates corresponds to the output of said down sampling means; and

face alignment means for aligning the face of the image output by said input means in accordance with said eye matching means;

feature extractor means for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration means; and

voting means for comparing the at least two extracted facial features with a database of corresponding facial features to identify the face;

wherein if said eye matching means does not determine that any one of the plurality of eye templates corresponds to the output of said down sample means, said down sampling means down samples at a second predetermined ratio the output signal from said edge extraction means.

70. A face recognition system comprising:

input means for inputting an image of a subject;

face detector means for determining if a face is present in the image output by said input means;

face position registration means for determining a position of the face in the image if said face detector means-determines that the face is present;

feature extractor means for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration means; and

voting means for comparing the at least two extracted facial features with a database of corresponding facial features to identify the face;

wherein said feature extraction means comprises;

cropping means for cropping at least the first facial feature and the second facial feature output by said face position registration means;

lumen normalization means for lumen normalizing the first facial feature and the second facial feature output by said cropping means;

edge extraction means for extracting edge information from the first facial feature and the second facial feature output by said lumen normalization means;

down sampling means for down sampling at a first predetermined ratio the first facial feature and at a second predetermined ratio the second facial feature output signal by said edge extraction means; and

feature vector calculating means for calculating a first feature vector of the first facial feature from said down sampling means and a second feature vector of the second facial feature output from said down sampling means.

71. A face recognition system according to claim 70, wherein the first predetermined ratio is greater than the second predetermined ratio.

72. A face recognition system according to claim 70,

wherein said cropping means crops a third facial output by said face position registration means;

wherein said lumen normalization means lumen normalizes the third facial feature output by said cropping means detector means;

wherein said edge extraction means extracts edge information from the third facial feature output by said lumen normalization means;

wherein said down sampling means down samples at a third predetermined ratio the third facial feature output by said extraction means; and

wherein said feature vector calculating means calculates a third feature vector of the third facial feature from said down sampling means.

73. A face recognition system according to claim 72, wherein the first predetermined ratio is greater than the second predetermined ratio, and wherein the second ratio is greater than the third ratio.

79. A face recognition system according to claim 70, wherein said voting means comprises

first recognizer means for comparing the first feature with corresponding first features of a database of candidates;

second recognizer means for comparing the second feature with corresponding second features of the database of candidates; and

selection means of selecting one of the candidates in the database in accordance with one of said first recognizer means and said second recognizer means.

74. A face recognition system comprising:

input means for inputting an image of a subject;

face detector means for determining if a face is present in the image output by said input means;

face position registration means for determining a position of the face in the image if said face detector means determines that the face is present;

feature extractor means for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration means; and

voting means for comparing the at least two extracted facial features with a database of corresponding facial features to identify the face;

wherein said feature extraction means comprises;

cropping means for cropping at least a first facial feature and a second facial feature output by said face position registration means;

down sampling means for down sampling at a first predetermined ratio the first facial feature and at a second predetermined ratio the second facial feature output signal by said cropping means; and

feature vector calculating means for calculating a first feature vector of the first facial feature from said down sampling means and a second feature vector of the second facial feature output from said down sampling means.

75. A face recognition system according to claim 74, wherein the first predetermined ratio is greater than the second predetermined ratio.

76. A face recognition system according to claim 74,

wherein said cropping means crops a third facial output by said face position registration means;

wherein said down sampling means down samples at a third predetermined ratio the third facial feature output by said cropping means; and

wherein said feature vector calculating means calculates a third feature vector of the third facial feature from said down sampling means.

77. A face recognition system according to claim 74, wherein said voting means comprises

first recognizer means for comparing the first feature vector with corresponding first feature vectors of a database of candidates;

second recognizer means for comparing the second feature vector with corresponding second feature vectors of the database of candidates; and

selection means of selecting one of the candidates in the database in accordance with one of said first recognizer means and said second recognizer means.

78. A face recognition system according to claim 76, wherein the first predetermined ratio is greater than the second predetermined ratio, and wherein the second ratio is greater than the third ratio.

81. A face recognition system comprising:

input means for inputting an image of a subject;

face position registration means for determining a position of a face in the image output from said input means, said face position registration means comprising eye template means for applying an eye template to a first side of the face and a mirror image of the eye template to a second side of the face;

feature extractor means for extracting at least first and second facial features from the face; and

voting means for comparing the first and second facial features extracted by said feature extractor means with a database of extracted facial features to identify the face.

82. A face recognition system according to claim 81, wherein said face position registration means comprises a neural network.

83. An article of manufacture, comprising a computer usable medium having readable code means embodied therein for face recognition, the computer readable program code means comprising:

input computer readable program code means for inputting an image of a subject;

face detector computer readable program code means for determining if a face is present in the image from said input computer readable program code means;

face position registration computer readable program code means for determining a position of the face in the image if said face detector computer readable program code means determines that the face is present, said face position registration computer readable program code means comprising computer readable program code means for applying an eye template to a first side of the face and a mirror image of the eye template to a second side of the face;

feature extractor computer readable program code means for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration computer readable program code means; and

voting computer readable program code means for comparing the at least two extracted facial features with a database of corresponding facial features to identify the face.

84. An article of manufacture according to claim 83, wherein said face detector computer readable program code mean comprises a neural network.

85. An article of manufacture according to claim 83, wherein said face position registration computer readable program code mean comprises a neural network.

86. An article of manufacture according to claim 83, wherein said feature extractor computer readable program code mean comprises a neural network.

87. A face recognition system comprising:

a control panel having an input circuit and a display;

a video camera for inputting an image of a subject in response to said control panel;

a face detector for determining if a face is present in the image output by said video camera, if the face is not present a message is displayed by said display;

a face position registration circuit for determining a position of the face in the image if said face detector determines that the face is present, said face position registration circuit comprising an eye template circuit that applies an eye template to a first side of the face and a mirror image of the eye template to a second side of the face;

a feature extractor for extracting at least first and second facial features from the face in accordance with the position determined by said face position registration circuit;

a face sever comprising a database of corresponding features of potential candidates; and

voting circuit for comparing the at least two extracted facial features with the database from said face server to identify the face.

88. A face recognition system according to claim 87, wherein said face detector comprises a neural network.

89. A face recognition system according to claim 87, wherein said face position registration circuit comprises a neural network.

90. A face recognition system according to claim 87, wherein said feature extractor comprises a neural network.