APPROACHES AND APPARATUS FOR EYE DETECTION IN A DIGITAL IMAGE

• US 20070189582A1
• Filed: 02/07/2007
• Published: 08/16/2007
• Est. Priority Date: 01/26/2005
• Status: Active Grant

• Current Assignee: Gentex Corporation
• Original Assignee: Honeywell International Inc.
• Inventors: Hamza, Rida, Ahrens, Terry
• Granted Patent: US 8,090,157 B2
• Time in Patent Office: Days
• US Class Current: 382/117
• 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

A system for finding and providing images of eyes acceptable for review, recordation, analysis, segmentation, mapping, normalization, feature extraction, encoding, storage, enrollment, indexing, matching, and/or the like. The system may acquire images of the candidates run them through a contrast filter. The images may be ranked and a number of candidates may be extracted for a list from where a candidate may be selected. Metrics of the eyes may be measured and their profiles evaluated. Also, the spacing between a pair of eyes may be evaluated to confirm the pair'"'"'s validity. Eye images that do not measure up to certain standards may be discarded and new ones may be selected.

Claims

1. A candidate finder system comprising:

a filter;

a first candidate selector connected to the filter;

a first profile validator connected to the candidate selector;

a first candidate eliminator connected to the second profile validator and the first candidate selector;

a second candidate selector connected to the first profile validator;

a pair validator connected to the second candidate determiner;

a second profile validator connected to the pair validator; and

a second candidate eliminator connected to the first profile validator and the second candidate determiner.

2. The system of claim 1, further comprising:

a resulter connected to the first and second profile validators; and

a stopper connected to the first and second candidate eliminators.

3. The system of claim 1, wherein a candidate is an eye.

4. The system of claim 1, wherein:

the first candidate selector comprises;

a ranking mechanism connected to the filter;

a candidates extractor connected to the ranking mechanism; and

a first candidate determiner connected to the candidate extractor, the first profiler and the first candidate eliminator; and

the second candidate selector comprises;

the ranking mechanism connected to the filter;

the candidates extractor connected to the ranking mechanism; and

a second candidate determiner connected to the first profile evaluator, the pair validator and the second candidate eliminator.

5. The system of claim 1, wherein:

the first profile validator comprises;

a first profiler connected to the first candidate selector; and

a first profile evaluator connected to the first profiler, the first candidate eliminator and the second candidate selector; and

the second profile validator comprises;

a second profiler connected to the pair validator; and

a second profile evaluator connected to the second profiler and the second candidate eliminator.

6. The system of claim 5, wherein:

a candidate is an eye having a pupil;

the first profiler comprises;

a pupil region extractor connected to the first candidate selector;

an adaptive thresholder connected to the pupil region extractor;

a contours finder connected to the adaptive thresholder;

a contour picker connected to the contours finder;

a curve fitter connected to the contour picker; and

a curve selector connected to the curve fitter and the first profile evaluator; and

the second profiler comprises;

a pupil region extractor connected to the pair validator candidate selector;

an adaptive thresholder connected to the pupil region extractor;

a contours finder connected to the adaptive thresholder;

a contour picker connected to the contours finder;

a curve fitter connected to the contour picker; and

a curve selector connected to the curve fitter and the second profile evaluator.

7. The system of claim 6, wherein an adaptive threshold is computed adaptively based upon a pupil size and a contrast distribution of an image region having a pupil and its surrounding area.

8. The system of claim 1, wherein the pair validator comprises:

a space measurer connected to the second candidate selector; and

a range indicator connected to the space measurer and the second profile validator.

9. The system of claim 1, further comprising:

a resulter connected to the first and second profile validators; and

wherein the resulter is for receiving a valid pair of candidates.

10. The system of claim 1, wherein:

the filter is a contrast filter; and

the candidates are eyes.

11. The system of claim 10, wherein the contrast filter is based upon a pupil dark profile with respect to its surroundings.

12. The system of claim 10, wherein the contrast filter is based upon a light reflection profile with respect to its surroundings

13. A method for eye finding, comprising:

providing candidates;

contrast filtering the candidates;

selecting a first eye from the candidates;

validating a profile of the first eye;

selecting a second eye from the candidates;

validating an amount of space between the first and second eyes; and

validating a profile of the second eye.

14. The method of claim 13, further comprising:

eliminating the first eye if the first eye has an invalid profile;

eliminating the second eye if the amount of space between the first and second eyes is invalid; and

eliminating the second eye if the amount of space between the first and second eyes is valid and the second eye has an invalid profile.

15. The method of claim 14, wherein:

the selecting a first eye comprises;

prioritizing the candidates according to contrast into a first list of candidates;

extracting a number of candidates from the first list of candidates into a second list of candidates; and

selecting a first eye from the second list of candidates; and

the selecting a second eye comprises selecting a second eye from the second list of candidates.

16. The method of claim 15, wherein:

the validating a profile of first eye comprises;

measuring profile metrics of the first eye; and

determining whether the profile metrics of the first eye are valid; and

the validating a profile of second eye comprises;

measuring profile metrics of the second eye; and

determining whether the profile metrics of the second eye are valid.

17. The method of claim 16, wherein the validating an amount of space between the first and second eyes comprises:

measuring a spacing between the first and second eyes; and

determining whether the spacing is within a valid range.

18. An eye finder system comprising:

a camera;

a filter connected to the camera;

a candidate lister connected to the filter;

an eye selector connected to the candidate lister; and

an eye profile evaluator connected to the eye selector.

19. The system of claim 18, further comprising an eye eliminator connected to the eye profile evaluator.

20. The system of claim 19, wherein:

the eye eliminator comprises;

a deleter, connected to the eye profile evaluator, for removing the eye from a list provided by the candidate lister if the eye profile evaluator indicates a profile of the eye to be invalid; and

a counter connected to the deleter and to the eye selector;

if the counter has a count greater than zero, then the eye selector may select a new eye from the list provided by the candidate lister;

or

if the counter has a count not greater than zero, then the system may stop.

21. The system of claim 18, further comprising:

a second eye selector connected to the eye profile evaluator;

an eye pair evaluator connected to the second eye selector; and

a second profile evaluator connected to the eye pair evaluator.

22. The system of claim 18, wherein the eye profile evaluator comprises:

a pupil region extractor connected to the eye selector;

an adaptive thresholder connected to the pupil region extractor;

a contours finder connected to the adaptive thresholder;

a contour picker connected to the contours finder;

a curve fitter connected to the contour picker; and

a best curve selector connected to the curve fitter.

23. The system of claim 21, wherein if the profile evaluator and the second profile evaluator indicate that a first eye and a second eye have appropriate metrics, and the eye pair evaluator determines that the first eye and second eye are appropriately spaced from each other, then the first and second eyes or portions of them are acceptable for review, recordation, analysis, segmentation, mapping, normalization, feature extraction, encoding, storage, enrollment, indexing, matching, and/or the like.

24. The system of claim 21, wherein if just one eye is indicated by one or more profile evaluators to have appropriate metrics, then the eye or a portion of it is acceptable for review, recordation, analysis, segmentation, mapping, normalization, feature extraction, encoding, storage, enrollment, indexing, matching, and/or the like.