SYSTEM AND METHOD FOR GENERATING AND EMPLOYING SHORT LENGTH IRIS CODES

• US 20110075892A1
• Filed: 09/25/2009
• Published: 03/31/2011
• Est. Priority Date: 09/25/2009
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Connell, Jonathan H. II, Ratha, Nalini K., Gentle, James E.
• Granted Patent: US 10,216,995 B2
• Time in Patent Office: Days
• US Class Current: 382/117
• CPC Class Codes: G06F 16/50 of still image data
• CPC Class Codes: G06F 16/51 Indexing; Data structures therefor; Storage structures
• CPC Class Codes: G06F 16/53 Querying
• CPC Class Codes: G06F 16/5838 using colour
• CPC Class Codes: G06F 18/2113 by ranking or filtering the set of features, e.g. using a measure of variance or of feature cross-correlation
• CPC Class Codes: G06T 2207/20112 Image segmentation details
• CPC Class Codes: G06T 2207/30196 Human being; Person
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06V 10/771 Feature selection, e.g. selecting representative features from a multi-dimensional feature space
• CPC Class Codes: G06V 40/197 Matching; Classification
• CPC Class Codes: G06V 40/40 Spoof detection, e.g. liveness detection

Abstract

A system and method for generating compact iris representations based on a database of iris images includes providing full-length iris codes for iris images in a database, where the full-length iris code includes a plurality of portions corresponding to circumferential rings in an associated iris image. Genuine and imposter score distributions are computed for the full-length iris codes, and code portions are identified that have a contribution that provides separation between imposter and genuine distributions relative to a threshold. A correlation between remaining code portions is measured. a subset of code portions having low correlations within the subset is generated to produce a compact iris representation.

Claims

1. A method for generating compact iris representations based on a database of iris images, the method comprising:

computing genuine and imposter score distributions for full-length iris codes for iris images in a database, where the full-length iris codes include a plurality of portions corresponding to circumferential rings in an associated iris image;

identifying and retaining code portions that have a contribution that provides separation between imposter and genuine distributions relative to a threshold;

measuring a correlation between remaining code portions; and

generating a subset of the remaining code portions having low correlations within the subset to produce a compact iris representation.

2. The method as recited in claim 1, further comprising determining parameters of a full length iris code to compact iris code transform by generating an all-aligned-pairs set.

3. The method as recited in claim 2, wherein identifying code portions includes computing a Hamming distance for all rows in an aligned code where a minimum Hamming distance is used as a criterion to analyze genuine and imposter score distributions for each row.

4. The method as recited in claim 1, wherein the separation is based upon a discriminatory capacity for rows in an iris code.

5. The method as recited in claim 4, wherein discriminatory capacity is measured by a Kolmogorov-Smirnov (KS) analysis.

6. The method as recited in claim 1, wherein generating a subset includes selecting rows in an iris code having highest discriminatory capacity for the subset.

7. The method as recited in claim 6, wherein discriminatory capacity is measured by a Kolmogorov-Smirnov (KS) analysis.

8. The method as recited in claim 1, wherein identifying code portions includes identifying code portions in rows of an iris code for both a real and imaginary portion of the iris code.

9. The method as recited in claim 1, wherein generating a subset of code portions includes subsampling rows in a region of interest between rows 10 and 52 of the iris code.

10. The method as recited in claim 1, wherein generating a subset includes computing a Hamming distance between two portions from a same iris image and taking an average of this distance over the iris images in the database.

11. A computer readable storage medium comprising a computer readable program for generating compact iris representations based on a database of iris images, wherein the computer readable program when executed on a computer causes the computer to perform the steps of

computing genuine and imposter score distributions for full-length iris codes for iris images in a database, where the full-length iris codes include a plurality of portions corresponding to circumferential rings in an associated iris image;

identifying and retaining code portions that have a contribution that provides separation between imposter and genuine distributions relative to a threshold;

measuring a correlation between remaining code portions; and

generating a subset of the remaining code portions having low correlations within the subset to produce a compact iris representation.

12. The computer readable storage medium as recited in claim 11, further comprising determining parameters of a full length iris code to compact iris code transform by generating an all-aligned-pairs set.

13. The computer readable storage medium as recited in claim 12, wherein identifying code portions includes computing a Hamming distance for all rows in an aligned code where a minimum Hamming distance is used as a criterion to analyze genuine and imposter score distributions for each row.

14. The computer readable storage medium as recited in claim 11, wherein the separation is based upon a discriminatory capacity for rows in an iris code.

15. The computer readable storage medium as recited in claim 14, wherein discriminatory capacity is measured by a Kolmogorov-Smirnov (KS) analysis.

16. The computer readable storage medium as recited in claim 11, wherein generating a subset includes selecting rows in an iris code having highest discriminatory capacity for the subset.

17. The computer readable storage medium as recited in claim 16, wherein discriminatory capacity is measured by a Kolmogorov-Smirnov (KS) analysis.

18. The computer readable storage medium as recited in claim 11, wherein identifying code portions includes identifying code portions in rows of an iris code for both a real and imaginary portion of the iris code.

19. The computer readable storage medium as recited in claim 11, wherein generating a subset of code portions includes subsampling rows in a region of interest between rows 10 and 52 of the iris code.

20. The computer readable storage medium as recited in claim 11, wherein generating a subset includes computing a Hamming distance between two portions from a same iris image and taking an average of this distance over the iris images in the database.

21. A system for generating compact iris representations based on a database of iris images comprising:

a database configured to store full-length iris codes for all iris images, where each iris code includes a plurality of portions corresponding to circumferential rings in an associated iris image; and

a processor configured to compute genuine and imposter score distributions for the portions of iris codes and determine which code portions of the iris codes provide a highest contribution to a separation between imposter and genuine distributions, the processor being configured to measure a correlation between the remaining code portions to determine a subset of the code portions having lowest correlations within the subset such that a compact iris representation is generated from the subset.

22. The system as recited in claim 21, further comprising a transform generated from the subset to generate compact iris codes.

23. The system as recited in claim 21, wherein the process executes program code for identifying highest contribution code portions by computing a Hamming distance for all rows in an aligned code where a minimum Hamming distance is used as a criterion to analyze genuine and imposter score distributions for each row;

and the subset of code portions includes a highest discriminatory capacity for rows in an iris code.

24. The system as recited in claim 23, wherein the separation is measured by a Kolmogorov-Smirnov (KS) analysis.

25. The system as recited in claim 21, wherein the code portions having a highest contribution includes rows of an iris code between rows 10 and 52 of both a real and imaginary portion of the iris code.