Iris recognition using hyper-spectral signatures

1. A system of iris recognition, comprising:

storage configured to store reference hyper-spectral signatures for different portions of the iris of a person's eye for each of a plurality of identified persons, each said portion comprising one or more spatial pixels at a certain spatial resolution, each said reference hyper-spectral signature defined by features extracted from data contained in four or more bands within the RGB color space from approximately 400 nm to 700 nm, said reference hyper-spectral signatures for the different portions of the iris combining spectral diversity of the four or more bands and spatial diversity of said portions at the certain spatial resolution to distinguish color signatures that are not visually distinguishable by the human eye;

a hyper-spectral sensor configured to sense data through an aperture at the certain spatial resolution from at least said four or more bands for the different portions of the iris in response to stimuli from an unidentified person's iris at a given range, the size of said aperture being at least ten times smaller than and providing at least ten times less certain spatial resolution than a near-infrared (NIR) sensor configured to produce a monochrome texture signature to identify the unidentified person at the same range; and

a classifier configured to extract features from the data contained in the four or more bands in the different portions of the iris to produce sensed hyper-spectral signatures of the different portions of the iris at the certain spatial resolution of the unidentified person and compare the sensed hyper-spectral signature to the reference hyper-spectral signature for the same portion of the iris for the plurality of different portions to generate an output that identifies the unidentified person.

2. The system of claim 1 , wherein both the reference and sensed hyper-spectral signatures includes six or more bands within the RGB color space from approximately 400 nm to 700 nm

3. The system of claim 1 , wherein the four or more bands are non-contigous.

4. The system of claim 1 , wherein the four or more bands include at least one band in each of the R, G and B bands.

5. The system of claim 1 , the hyper-spectral signatures further comprising at least one band in one of the UV and NIR bands.

6. The system of claim 1 , wherein the classifier generates an output for each said portion, further comprising a fusion classifier that fuses the outputs to identify the unidentified person.

7. The system of claim 1 , further comprising a training pre-processor that uses hyper-spectral signatures for a plurality of people as a training set to down select the four or more bands from a larger plurality of candidate bands within the RGB color space from approximately 400 nm to 700 nm to distinguish the hyper-spectral signatures in the training set.

8. The system of claim 1 , wherein the sensor measures a signature of the sclera of the unidentified person's eye as a calibration reference.

9. The system of claim 1 , wherein measurement of the hyper-spectral signature is the only basis for iris recognition of the unidentified person.

10. The system of claim 1 , wherein the classifier processes hyper-spectral signatures sensed at said given range in excess of 25 m to identify the unidentified person.

11. A method of iris recognition, comprising:

storing reference hyper-spectral signatures for different portions of the iris of a person's eye for each of a plurality of identified persons, each said portion comprising one or more spatial pixels at a certain spatial resolution, each said reference hyper-spectral signature defined by features extracted from data contained in four or more bands within the RGB color space from approximately 400 nm to 700 nm, said reference hyper-spectral signatures for the different portions of the iris combining spectral diversity of the four or more bands and spatial diversity of said portions at the certain spatial resolution to distinguish color signatures that are not visually distinguishable by the human eye;

sensing data with a hyper-spectral sensor through an aperture at the certain spatial resolution from at least said four or more bands for the different portions of the iris in response to stimuli from an unidentified person's iris at a given range, the size of said aperture being at least ten times smaller than and providing at least ten times less certain spatial resolution than a near-infrared (NIR) sensor configured to produce a monochrome texture signature to identify the unidentified person at the same range; and

extracting features from the data contained in the four or more bands in the different portions of the iris to produce sensed hyper-spectral signatures of the different portions of the iris at the certain spatial resolution of the unidentified person and comparing the sensed hyper-spectral signature to the reference hyper-spectral signature for the same portion of the iris for the plurality of different portions to generate an output that identifies the unidentified person.

12. The method of claim 11 , wherein both the reference and sensed hyper-spectral signatures includes at least six non-contiguous bands.

13. The method of claim 11 , wherein both the reference and sensed hyper-spectral signatures include six or more bands within the RGB color space from approximately 400 nm to 700 nm.

14. The method of claim 11 , further comprising:

providing hyper-spectral signatures for a plurality of people as a training set;

providing a larger plurality of candidate bands within the RGB color space from approximately 400 nm to 700 nm; and

using the training set to down select the four or more bands from the larger plurality of candidate bands to distinguish the hyper-spectral signatures in the training set.

15. The method of claim 11 , wherein said four or more bands include at least two bands within at least one of the R band, the G band or the B band.

16. The system of claim 1 , wherein said four or more bands include at least two bands within at least one of the R band, the G band or the B band.