Image based multi-biometric system and method

• US 7,697,735 B2
• Filed: 06/21/2005
• Issued: 04/13/2010
• Est. Priority Date: 06/21/2004
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Adam, Hartwig, Steffens, Johannes B., Neven, Hartmut
• Primary Examiner(s): PERUNGAVOOR, SATHYANARAYA V
• Application Number:
  US11/158,906
• Publication Number:
  US 20060050933A1
• Time in Patent Office: 1,757 Days
• Field of Search: 382/118
• US Class Current: 382/118
• CPC Class Codes: G06F 18/256 of results relating to different input data, e.g. multimodal recognition
• CPC Class Codes: G06V 10/811 the classifiers operating on different input data, e.g. multi-modal recognition
• CPC Class Codes: G06V 40/10 Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
• CPC Class Codes: G06V 40/16 Human faces, e.g. facial parts, sketches or expressions
• CPC Class Codes: G06V 40/171 Local features and components; Facial parts eye characteristics G06V40/18; Occluding parts, e.g. glasses; Geometrical relationships
• CPC Class Codes: G06V 40/18 Eye characteristics, e.g. of the iris
• CPC Class Codes: G06V 40/193 Preprocessing; Feature extraction

Abstract

This disclosure describes methods to integrate face, skin and iris recognition to provide a biometric system for identifying individuals. These methods only require a digital image depicting a human face as source data.

Claims

1. A face recognition method, the method performed by a computer and comprising:

selecting a single image of a face;

locating facial features in the single image;

determining whether a plurality of the located facial features lie within a plane;

responsive at least in part to determining whether the plurality of the located facial features lie within a plane, determining whether the single image represents a recorded image of an individual;

responsive at least in part to determining that the single image does not represent a recorded image of an individual;

comparing facial features located in the single image with facial features of a reference image;

comparing iris features in the single image with iris features in the reference image;

comparing skin characteristics in the single image with skin characteristics in the reference image, the comparison of skin characteristics comprising;

locating skin areas in the face suitable for skin texture analysis;

responsive to locating the facial features, detecting deformation of the located skin areas;

responsive to detecting the deformation, warping, aligning, and normalizing the located skin areas to account for the deformation, resulting in modified skin areas; and

extracting skin characteristics from the modified skin areas;

calculating a similarity score between the single image and the reference image based on the facial feature comparison, the iris feature comparison, and the skin characteristics comparison; and

recognizing the face based at least in part on the calculated similarity score.

2. The face recognition method of claim 1, wherein the single image has an eye distance resolution of at least 600 pixels.

3. The face recognition method of claim 1, wherein the single image comprises at least 5 mega pixels.

4. The face recognition method of claim 1, wherein the skin characteristics consist of at least one of pores, wrinkles, and moles.

5. The face recognition method of claim 1, wherein the facial features are located using elastic bunch graph matching.

6. The face recognition method of claim 1, wherein the facial features are located using Gabor wavelets.

7. A face recognition apparatus, comprising:

means for selecting a first image of a face;

means for locating a first set of facial features in the first image;

means for selecting a second image of the face;

means for locating a second set of facial features in the second image, the second set of facial features corresponding to the facial features of the first set of facial features;

means for measuring a degree of change between the diameter of a pupil of an eye of the first image and the diameter of a pupil of an eye of the second image;

means for determining that the first image represents a real-time image of a living individual, responsive at least in part to the measured degree of change;

means for comparing the first set of facial features with facial features of a reference image, responsive at least in part to determining that the single image represents a real-time image of a living individual;

means for comparing iris features in the first image with iris features in the reference image, responsive at least in part to determining that the single image represents a real-time image of a living individual;

means for comparing skin characteristics of the first image with skin characteristics of the reference image, the comparison of skin characteristics comprising;

locating skin areas in the first image suitable for skin texture analysis;

responsive to locating the first set of facial features, detecting deformation of the located skin areas;

responsive to detecting the deformation, warping, aligning, and normalizing the located skin areas to account for the deformation, resulting in modified skin areas; and

extracting skin characteristics from the modified skin areas;

means for calculating a similarity score between the single image and the reference image based on the facial feature comparison, the iris feature comparison, and the skin characteristics comparison, responsive at least in part to determining that the single image represents a real-time image of a living individual; and

means for recognizing the face based at least in part on the calculated similarity score.

8. The face recognition apparatus of claim 7, wherein the first image has an eye distance resolution of at least 600 pixels.

9. The face recognition apparatus of claim 7, wherein the first image comprises at least 5 mega pixels.

10. The face recognition apparatus of claim 7, wherein the skin characteristics consist of at least one of pores, wrinkles, and moles.

11. The face recognition apparatus of claim 7, wherein the means for determining that the single image represents a real-time image of a living individual is further responsive at least in part to analyzing internal facial movements between the first set of facial features and the second set of facial features.

12. A computer readable storage medium storing a computer program executable by a processor for performing face recognition, the actions of the computer program comprising:

selecting a single image of a face;

locating facial features in the single image;

determining whether a plurality of the located facial features lie within a plane;

responsive at least in part to determining whether the plurality of the located facial features lie within a plane, determining whether the single image represents a recorded image of an individual;

responsive at least in part to determining that the single image does not represent a recorded image of an individual;

comparing facial features located in the single image with facial features of a reference image;

comparing iris features in the single image with iris features in the reference image;

comparing skin characteristics in the single image with skin characteristics in the reference image, the comparison of skin characteristics comprising;

locating skin areas in the face suitable for skin texture analysis;

responsive to locating the facial features, detecting deformation of the located skin areas;

responsive to detecting the deformation, warping, aligning, and normalizing the located skin areas to account for the deformation, resulting in modified skin areas; and

extracting skin characteristics from the modified skin areas;

calculating a similarity score between the single image and the reference image based on the facial feature comparison, the iris feature comparison, and the skin characteristics comparison; and

recognizing the face based at least in part on the calculated similarity score.

13. The computer readable storage medium of claim 12, wherein the single image has an eye distance resolution of at least 600 pixels.

14. The computer readable storage medium of claim 12, wherein the single image comprises at least 5 mega pixels.

15. The computer readable storage medium of claim 12, wherein the skin characteristics consist of at least one of pores, wrinkles, and moles.

16. The computer readable storage medium of claim 12, wherein the facial features are located using elastic bunch graph matching.

17. The computer readable storage medium of claim 12, wherein the facial features are located using Gabor wavelets.