Hybrid optical-digital method and apparatus for fingerprint verification

• US 5,761,330 A
• Filed: 06/07/1995
• Issued: 06/02/1998
• Est. Priority Date: 06/07/1995
• Status: Expired due to Term

• Current Assignee: Bioscrypt, Inc. (Advent International Corporation)
• Original Assignee: Bioscrypt, Inc. (Advent International Corporation)
• Inventors: Stoianov, Alexei, Soutar, Colin, Tomko, George J.
• Primary Examiner(s): Moore, David K.
• Assistant Examiner(s): Werner, Brian P.
• Application Number:
  US08/482,404
• Time in Patent Office: 1,091 Days
• Field of Search: 382/124, 382/280, 382/116, 382/125, 382/126, 382/127, 356/71
• US Class Current: 382/127
• CPC Class Codes: G06V 40/12 Fingerprints or palmprints

Abstract

A hybrid optical-digital technique is provided for automatic fingerprint verification. A coherent optical beam modulated with the characteristics of a fingerprint image passes through the lens performing an optical Fourier transform. The intensity array that represents the Fourier power spectrum is recorded and processed. In enrolment, several arrays related to the same finger are captured and processed. A template which is unique for each fingerprint is generated and stored. Using the set of arrays captured in enrolment and a set of templates from a standard database, the processor calculates dependences of false rejection rate and false acceptance rate upon a metric of comparison. The metric is obtained by calculating a relative distance between two arrays in Hilbert space. An individual threshold of verification for a person to be enroled is set and stored together with the template. In verification, the metric of comparison between the processed array derived from a fingerprint to be verified and the pre-stored template is determined and compared with the pre-stored individual threshold.

Claims

1. A hybrid optical/digital fingerprint enrolment apparatus comprising:

a source of coherent light for providing an illuminating beam along a beam path;

an input in said beam path for producing an optical signal carrying information about a fingerprint of a user along an optical signal path;

optical Fourier transform apparatus in said optical signal path for obtaining a Fourier spatial spectrum of said optical signal;

means for obtaining a digitised distribution related to a two-dimensional intensity distribution of said Fourier spatial spectrum;

means for storing said intensity-related distribution as one of a set of intensity-related distributions for said user;

means for storing a set of comparator template distributions;

means for deriving a fingerprint template distribution from said set of intensity-related distributions;

means for choosing at least two intensity-related distributions in said set of intensity-related distributions;

means for calculating a metric of comparison between each chosen intensity-related distribution and at least two comparator template distributions in said set of comparator template distributions;

means for calculating said metric of comparison between each distribution in said set of intensity-related distributions which was not used for generation of the fingerprint template distribution and said fingerprint template distribution;

means, responsive to said means for calculating a metric of comparison between each chosen integrity-related distribution and at least two comparator template distributions and responsive to said means for calculating said metric of comparison between each distribution in said set of intensity-related distributions which was not used for generation of the fingerprint template distribution and said fingerprint template distribution, for deriving a user dependent threshold of verification; and

means for storing said fingerprint template distribution and said user dependent threshold of verification.

2. The apparatus of claim 1 wherein said means for calculating a metric of comparison between each chosen intensity-related distribution and at least two comparator template distributions in said set of comparator template distributions is for determining a false acceptance rate versus said metric and wherein said means for calculating said metric of comparison between each distribution in said set of intensity-related distributions which was not used for generation of the fingerprint template distribution, and said fingerprint template distribution and is for determining a false rejection rate versus said metric such that said means for deriving a user dependent threshold of verification is based on said false acceptance rate and said false rejection rate.

3. A hybrid optical/digital method of fingerprint enrolment comprising the steps of:

(a) obtaining an optical signal modulated with characteristics of a user'"'"'s fingerprint;

(b) obtaining an optical Fourier spectrum of said optical signal;

(c) obtaining an array related to a two-dimensional intensity distribution of said Fourier spectrum;

(d) repeating steps (a) to (c) at least twice for said user'"'"'s fingerprint in order to obtain a set of intensity-related arrays;

(e) retrieving a set of comparator template arrays;

(f) choosing at least two intensity-related arrays in said set of intensity-related array and each comparator template array in said set of comparator template arrays;

(g) calculating a metric of comparison between each chosen intensity-related array and each comparator template array in said set of comparator template arrays;

(h) generating a fingerprint template array from said set of intensity-related arrays;

(i) calculating said metric of comparison between each array in said set of intensity-related arrays which were not used for generation of the fingerprint template array and said fingerprint template array;

(j) determining a user dependent threshold of verification based on step (g) and step (i), and

(k) storing said user dependent threshold of verification.

4. The method of claim 3 wherein step (c) comprises the steps of:

filtering said Fourier spectrum;

registering an intensity array related to a two-dimensional intensity distribution of said filtered Fourier spectrum;

mapping said registered intensity array, comprising the steps of;

(i) determining an average intensity over said registered intensity array;

(ii) dividing each element in said registered intensity array by said average intensity to obtain a normalised intensity array; and

(iii) mapping said normalised intensity array.

5. The method of claim 4 wherein the step of mapping said normalised intensity array comprises mapping said normalised intensity array with a saturation function.

6. The method of claim 4 wherein the step of mapping said normalised intensity array comprises mapping said normalised intensity array with a binarising function.

7. The method of claim 3 wherein calculating said metric of comparison comprises calculating a relative distance between said set of intensity-related arrays and one of said fingerprint template array and said comparator template array in Hilbert space.

8. The method of claim 3 wherein step (c) includes the steps of:

filtering said Fourier spectrum;

registering an intensity array related to a two-dimensional intensity distribution of said filtered Fourier spectrum;

mapping said registered intensity array, comprising the steps of;

(i) selecting a number of pixels of said array to generate a selected pixels intensity array;

(ii) determining an average intensity over said selected pixels intensity array;

(iii) dividing each element in said selected pixels intensity array by said average intensity to obtain a normalised selected pixels intensity array;

(iv) mapping said normalised selected pixels intensity array into a mapped intensity array;

(v) transforming said mapped intensity array to a polar co-ordinate system, the centre of which corresponds to the zeroth spatial frequency of said Fourier spectrum, to generate a polar co-ordinates intensity array, each element in said polar co-ordinates intensity array being represented by a vector comprising polar co-ordinates and a mapped intensity.

9. The method of claim 8 wherein calculating said metric of comparison comprises the steps of:

for each array in said set of intensity-related arrays which have not been used for generation of the fingerprint template array, applying a function to calculate a difference between each vector in said each array and a corresponding vector in one of said fingerprint template array and said comparator template array;

plotting each said difference in a histogram;

scanning said histogram for a cluster of plotted differences which has a maximum number of plotted differences inside a space of pre-determined size; and

dividing the number of plotted differences within said cluster by the number of vectors in said one template array.

10. The method of claim 3 wherein said determining a user dependent threshold of comparison comprises the steps of:

determining a minimum false rejection rate;

determining a maximum false acceptance rate;

if said minimum false rejection rate exceeds said maximum false acceptance rate, choosing said user dependent threshold of comparison between said maximum false acceptance rate and said minimum false rejection rate, otherwise, determining a point of intersection of said false rejection rate versus said metric and false acceptance rate versus said metric and choosing said user dependent threshold of comparison proximate said point of intersection.

11. The method of claim 3 wherein the step (g) comprises determining a false acceptance rate versus said metric and wherein step (i) comprises determining a false rejection rate versus said metric such that the step of determining a user dependent threshold of verification is based on said false acceptance rate and said false rejection rate.

12. The method of claim 11 wherein said determining a false rejection rate versus said metric comprises:

sorting metrics obtained in step (i) in ascending order to generate a set of sorted metrics, each metric in said sorted set having an associated order number;

for each metric in said sorted set, dividing said order number by the number of metrics in said sorted set.

13. The method of claim 11 wherein said determining a false acceptance rate versus said metric comprises:

sorting metrics obtained in step (i) in descending order to generate a set of sorted metrics, each metric in said sorted set having an associated order number;

for each metric in said sorted set, dividing said order number by the number of metrics in said sorted set.

14. A method of fingerprint verification comprising the steps of:

(a) obtaining an optical signal modulated with characteristics of a user'"'"'s fingerprint;

(b) obtaining an optical Fourier spectrum of said optical signal;

(c) obtaining an array related to a two-dimensional intensity distribution of said Fourier spectrum including the steps of;

filtering said Fourier spectrum;

registering an intensity array related to a two-dimensional distribution of said Fourier spectrum;

mapping an intensity array, comprising the steps of;

(i) selecting a number of pixels of said array to generate a selected pixels intensity array;

(ii) determining an average intensity over said selected pixels intensity array;

(iii) dividing each element in said selected pixels intensity array by said average intensity to obtain a normalised selected pixels intensity array;

(iv) mapping said normalised selected pixels intensity array into a mapped intensity array;

(v) transforming said mapped intensity array to a polar co-ordinate system, the centre of which corresponds to the zeroth spatial frequency of said Fourier spectrum, to generate a polar co-ordinates intensity array, each element in said polar co-ordinates intensity array being represented by a vector comprising polar co-ordinates and a mapped intensity;

(d) retrieving fingerprint comparison data related to said fingerprint from a data store, said fingerprint comparison data comprising a fingerprint template array and a user dependent threshold of verification;

(e) indicating one of a verification and a non-verification responsive to said intensity-related array and said fingerprint comparison data including deriving a metric of comparison between said fingerprint template array-and said intensity-related array and comparing said user dependent threshold of verification with said metric.

15. The method of claim 14 wherein calculating said metric of comparison comprises the steps of:

applying a function to calculate a difference between each vector in said intensity-related array and a corresponding vector in said fingerprint template array;

plotting each said difference in a histogram;

scanning said histogram for a cluster of plotted differences which has a maximum number of plotted differences inside a space of pre-determined size; and

dividing the number of plotted differences within said cluster by the number of vectors in said fingerprint template array.

16. A hybrid optical/digital fingerprint enrolment and verification system comprising:

enrolment apparatus comprising;

a source of coherent light for providing an illuminating beam along an enrolment beam path;

an input in said enrolment beam path for producing an enrolment optical signal carrying information about a fingerprint of a user along an enrolment optical signal path;

optical Fourier transform apparatus in said enrolment optical signal path for obtaining a Fourier spatial spectrum of said enrolment optical signal;

means for obtaining a digitised enrolment distribution related to a two-dimensional intensity distribution of said Fourier spatial spectrum;

means for storing said intensity-related enrolment distribution as one of a set of intensity-related enrolment distributions for said user;

means for storing a set of comparator template distributions;

means for deriving a fingerprint template distribution from said set of intensity-related enrolment distributions;

means for choosing at least two intensity-related enrolment distributions in said set of intensity-related enrolment distributions;

means for calculating an enrolment metric of comparison between each chosen intensity-related enrolment distribution and at least two comparator template distributions in said set of comparator template distributions;

means for calculating said enrolment metric of comparison between each distribution in said set of intensity-related distributions which was not used for generation of the fingerprint template distribution and said fingerprint template distribution;

means, responsive to said means for calculating an enrolment metric of comparison between each chosen intensity-related enrolment distribution and at least two comparator template distributions and responsive to said means for calculating said enrolment metric of comparison between each distribution in said set of intensity-related distributions which was not used for generation of the fingerprint template distribution, and said fingerprint template distribution for deriving a user dependent threshold of verification; and

storage means for storing said fingerprint template distribution and said user dependent threshold of verification; and

verification apparatus, comprising;

a source of coherent light for providing an illuminating beam along a verification beam path;

an input in said verification beam path for producing a verification optical signal carrying information about a fingerprint along a verification optical signal path;

optical Fourier transform apparatus in said verification optical signal path for obtaining a verification Fourier spatial spectrum of said verification optical signal;

means for obtaining a verification digitised distribution related to a two-dimensional intensity distribution of said verification Fourier spatial spectrum;

means for retrieving said fingerprint template distribution from storage;

means for calculating a verification metric of comparison between said obtained verification digitised distribution and said fingerprint template distribution;

means for retrieving said user dependent threshold of verification from storage;

means for comparing said verification metric of comparison with said user dependent threshold of verification; and

means for indicating one of a verification and a non-verification responsive to said means for comparing.

17. The apparatus of claim 16 wherein said means for calculating an enrolment metric of comparison between each chosen intensity-related distribution and at least two comparator template distributions in said set of comparator template distributions is for determining a false acceptance rate versus said enrolment metric and wherein said means for calculating said metric of comparison between each distribution in said set of intensity-related distributions which was not used for generation of the fingerprint template distribution and said fingerprint template distribution and is for determining a false rejection rate versus said metric such that said means for deriving a user dependent threshold of verification is based on said false acceptance rate and said false rejection rate.

18. A hybrid digital-optical method of fingerprint enrolment and verification comprising:

(a) obtaining an optical signal modulated with characteristics of a user'"'"'s fingerprint;

(b) obtaining an optical Fourier spectrum of said optical signal;

(c) obtaining an array related to a two-dimensional intensity distribution of said Fourier spectrum;

(d) repeating steps (a) to (c) at least twice for said user'"'"'s fingerprint in order to obtain a set of intensity-related arrays;

(e) generating a fingerprint template array from said set of intensity-related arrays;

(f) retrieving a set of comparator template arrays;

(g) choosing at least two intensity-related arrays in said set of intensity-related arrays;

(h) calculating a metric of comparison between each chosen intensity-related array and each comparator template array in said set of comparator template arrays;

(i) calculating said metric of comparison between each array in said set of intensity-related arrays which were not used for generation of the fingerprint template array and said fingerprint template array;

(j) determining a user dependent threshold of verification based on step (h) and step (i);

(k) storing said user dependent threshold of verification and said fingerprint template array;

(l) obtaining a verification optical signal modulated with characteristics of a user'"'"'s fingerprint;

(m) obtaining a verification optical Fourier spectrum of said verification optical signal;

(n) obtaining a verification array related to a two-dimensional intensity distribution of said verification Fourier spectrum;

(o) retrieving said fingerprint template array and said user dependent threshold of verification; and

(p) deriving a verification metric of comparison between said fingerprint template array and said intensity-related verification array and comparing said user dependent threshold of verification with said verification metric and indicating one of a verification and a non-verification.