STANDOFF IRIS RECOGNITION SYSTEM

• US 20070140531A1
• Filed: 02/15/2007
• Published: 06/21/2007
• Est. Priority Date: 01/26/2005
• Status: Active Grant

• Current Assignee: Gentex Corporation
• Original Assignee: Honeywell International Inc.
• Inventors: Hamza, Rida
• Granted Patent: US 8,098,901 B2
• Time in Patent Office: Days
• US Class Current: 382/117
• CPC Class Codes: G06V 40/18 Eye characteristics, e.g. of the iris
• CPC Class Codes: G06V 40/193 Preprocessing; Feature extraction

Abstract

An iris recognition system having pupil and iris border conditioning prior to iris mapping and analysis. The system may obtain and filter an image of an eye. A pupil of the mage may be selected and segmented. Portions of the pupil border can be evaluated and pruned. A curve may be fitted on at least the invalid portions of the pupil border. The iris of the eye with an acceptable border of the pupil as an inside border of the iris may be selected from the image. The iris outside border having sclera and eyelash/lid boundaries may be grouped using a cluster angular range based on eye symmetry. The sclera boundaries may be fitted with a curve. The eyelash/lid boundaries may be extracted or masked. The iris may be segmented, mapped and analyzed.

Claims

1. An iris recognition system comprising:

an eyefinder;

a filter connected to the eyefinder;

a segmenter connected to the filter;

an edge validity evaluator connected to the segmenter; and

a curve fitter connected to the edge validity evaluator.

2. The system of claim 1, wherein:

the eyefinder is for providing a valid eye image; and

the filter is for smoothing out edges in the eye image.

3. The system of claim 1, further comprising a kernel module, for selecting an image of a pupil in the eye image, connected to the filter.

4. The system of claim 3, further comprising:

a curve implementation module connected to the curve fitter; and

wherein the curve implementation module is for establishing a border for the pupil in the image of the pupil, based on a curve from the curve fitter.

5. The system of claim 1, further comprising:

an active contour plus a curve implementation module connected to the curve fitter and to the segmenter; and

wherein the active contour plus a curve implementation module is for establishing a border in place of invalid edges of the pupil in the image of the pupil, based on a curve from the curve fitter, and for establishing a border from valid edges of the pupil.

6. The system of claim 5, further comprising a contour selector connected to the active contour plus a curve implementation module and connected to the curve fitter.

7. The system of claim 1, further comprising:

an edge pruner connected between the evaluator and the curve fitter; and

wherein the segmenter is a one dimensional polar segmenter.

8. An iris recognition system comprising:

an eyefinder;

a filter connected to the eyefinder;

a range module connected to the filter;

a segmenter connected to the range determiner;

a border module connected to the segmenter;

a count module connected to the border module; and

a curve fitter connected to the count module.

9. The system of claim 8, wherein:

the eyefinder is for providing a valid eye image having a processed pupil border; and

the filter is for smoothing out edges in the eye image.

10. The system of claim 8, further comprising a kernel module, for selecting an image of an iris in the eye image, connected to the filter.

11. The system of claim 8, further comprising a map analysis module connected to the count module, the curve fitter and the segmenter.

12. The system of claim 11, further comprising:

an extract eyelash/lid module connected to the count module and the map analysis module; and

a mask eyelash/lid module connected to the count module and the map analysis module.

13. The system of claim 8, wherein:

the range module is for setting a cluster angular range; and

the segmenter is a one dimensional polar plus segmentation module.

14. The system of claim 9, wherein:

the border module comprises a sclera border module and an eyelash/lid border module;

the count module is for determining a number of discontinuities in sclera borders;

the count module is for determining a number of discontinuities in the eyelash/lid borders;

if the number of discontinuities in the sclera borders is less than a first threshold, then the curve fitter is activated for curve fitting the sclera borders; and

if the number of discontinuities in the sclera borders is not less than the first threshold, then the eye image is invalid.

15. The system of claim 14, wherein:

if the number of discontinuities in the eyelash/lid borders is greater than a second threshold, then the eyelash/lid borders are masked; and

if the number of discontinuities in the eyelash/lid borders is not greater than the second threshold, then the eyelash/lid borders are extracted.

16. The system of claim 13, wherein the segmenter comprises:

a first one dimensional polar segmenter, for sclera borders, connected to the range module;

a second one dimensional polar segmenter, for eyelash/lid borders, connected to the range module;

a first get max peak module connected to the first one dimensional polar segmenter;

a second get max peak module connected to the second one dimensional polar segmenter;

a first one dimensional median filter connected to the first get max peak module and to the border module; and

a second one dimensional median filter connected to the second get max peak module and to the border module.

17. A method for iris recognition comprising:

providing an image of an eye;

selecting a pupil in the image;

segmenting the pupil;

determining a validity of portions of a border of the pupil; and

fitting a curve on at least invalid portions of the border of the pupil to form a resulting border of the pupil.

18. The method of claim 17, further comprising:

selecting an iris with the pupil having the resulting border from the image of the eye; and

clustering iris sclera boundaries and the eyelash/lid boundaries of the iris into first and second groups of boundaries, respectively.

19. The method of claim 18, further comprising:

determining a first number of discontinuities of the first group of boundaries; and

wherein;

if the first number is less than a first threshold, then the first group of boundaries is fitted with a curve fitting model; and

if the first number is not less than the first threshold, then the eye image is invalid.

20. The method of claim 19, further comprising:

determining a second number of discontinuities of the second group of boundaries; and

wherein;

if the second number is not greater than a second threshold, then the second group of boundaries is extracted;

if the second number is not greater than the second threshold and an area between outer borders of the second group of boundaries and an inner border of the iris is less than a third threshold, then the second group of boundaries are weighted accordingly; and

if the second number is greater than the second threshold, then the second group of boundaries is masked; and

further comprising mapping the iris.

21. The method of claim 19, wherein an iris map is constructed based upon actual inner and outer edge estimates with respect to fitting models.

22. The method of claim 21, wherein pixels of the iris map are extracted based upon an interpolation of image pixels within inner fitting model and outer fitting model edges at nearly all angles, deemed to be valid, with respect to a pupil center.

23. The method of claim 21, wherein nearly any pixel that lies within an outer border of the iris and a fitting model shape may be masked.

24. The method of claim 21, wherein nearly any pixel that lies outside an outer fitting model shape may be discarded.