Fingerprint identification system

• US 4,752,966 A
• Filed: 08/06/1986
• Issued: 06/21/1988
• Est. Priority Date: 03/26/1982
• Status: Expired due to Term

• Current Assignee: Fujifilm Electronic Imaging Ltd. (Fujifilm Holdings Corporation)
• Original Assignee: Secure Technologies Group, Inc.
• Inventors: Schiller, Michael
• Primary Examiner(s): Boudreau, Leo H.
• Application Number:
  US06/894,514
• Time in Patent Office: 685 Days
• Field of Search: 382/4, 382/5, 356/71
• US Class Current: 382/125
• CPC Class Codes: G06V 40/1347 Preprocessing; Feature extraction
• CPC Class Codes: G07C 9/37 using biometric data, e.g. fingerprints, iris scans or voice recognition

Abstract

An automatic fingerprint enrollment and identification technique is provided for automatically processing an input image represented by pixels having either a one or zero value. The automatic processing involves a series of steps in which the initial processing is a rough screening technique for eliminating a reasonable portion of the noise while passing through almost all of the minutia. As the steps proceed, the processing becomes more complex to provide a tighter screening out of noise while passing through as much of the minutia as possible. The later steps by operating on relatively fewer minutia or anomaly points that the earlier steps can be economical though more complex so as to be more discriminate than the earlier steps. The processing provides an optimum trade off between enrolling as large a proportion of the population as possible while minimizing the number of false identification errors.

Claims

1. In the machine method of identifying fingerprint minutia from a binary field of pixels having first and second binary signal values representing the fingerprint, the improvement comprising the steps of:

providing a set of binary value signals representing pixels of a fingerprint image generated from an optical scan of a fingerprint,

scanning each of a predetermined set of anomaly templates across the field of said pixels represented by said set of signals,

identifying as an anomaly each match of any one of said anomaly templates against a subfield of said pixels to provide a set of anomalies, each of said anomaly templates being of sufficient size to provide an indication of anomaly type, anomaly curvature and anomaly thickness,

testing the zone of pixels of contrasting value around each of said identified anomalies for thickness uniformity,

identifying as a minutia candidate those anomalies where the surrounding zone exhibits a uniformity of thickness in accordance with a predetermined criteria to provide a set of minutia candidates,

erecting a predetermined window around each of said set of minutia candidates,

identifying each of said anomalies in said window,

comparing the relationship between the set of said anomalies in said window against a predetermined fingerprint noise characteristic to provide a determination of the presence of fingerprint noise, and

rejecting the minutia candidate of said window where there has been a determination of fingerprint noise, to thereby provide a reduced set of minutia candidates.

2. The method of claim 1 further comprising the steps of:

establishing an analytical zone around each of said minutia candidates in said reduced set,

within each of said analytical zones comparing all anomalies except said minutia candidate to identify paired anomalies,

within said analytical zone, comparing said minutia candidate with each unpaired anomaly to identify any minutia candidate anomaly pair, and

eliminating from said reduced set of minutia candidates any of said minutia candidate anomaly pairs, to provide a set of minutia.

3. The method of claim 2 further comprising the steps of:

comparing each of a predetermined set of minutia angle identification templates against a representation of each of said set of minutia, each of said minutia angle identification templates incorporating the line ending and contrasting binary value bifurcation combination that defines a minutia,

correlating each of said angle identification templates to each minutia, and

assigning an angle to each minutia as a function of the template having the greatest correlation with the minutia involved.

4. The method of claim 1 wherein each member of said set of anomaly templates is at least three pixels by eight pixels.

5. The method of claim 1 wherein each of said anomaly templates is positioned over successive pixels so that overlapping subfields of said pixels are compared against a predetermined template to provide identification of anomalies.

6. The method of claim 1 wherein said step of testing comprises:

counting the distances from first and second predetermined points of each of said anomaly subfields in predetermined directions to predetermined transitions between the values in said field of pixels to provide first and second distance counts, and

comparing said first and second counts and eliminating from further consideration those anomalies wherein said comparison is past a predetermined threshold.

7. The method of claim 1 wherein said predetermined fingerprint noise characteristic is at least partially based on the number of anomalies in said window.

8. The method of claim 2 wherein said analytic zone and said window are the same.

9. The method of claim 3 wherein said normalized minutia is a skeletal minutia.

10. The method of claim 9 wherein said step of comparing includes a separate comparison of said angle indentification templates against said minutia at a plurality of laterally displaced positions for each of said angle identification templates.

11. The method of claim 3 wherein each of said set of anomaly templates is three pixels by eight pixels.

12. The method of claim 10 wherein each of said anomaly templates is positioned over successive pixels so that overlapping subfields of said pixels are compared against a predetermined template to provide identification of anomalies.

13. The method of claim 12 wherein said step of testing comprises:

counting the distances from first and second predetermined points of each of said anomaly subfields in predetermined directions to predetermined transitions between the values in said field of pixels to provide first and second distance counts, and

comparing said first and second counts and eliminating from further consideration those anomalies wherein said comparision is past a predetermined threshold.

14. The method of claim 13 wherein said predetermined fingerprint noise characteristic is at least partially based on the number of anomalies in the said window.

15. The method of claim 14 wherein said analytic zone and said window are the same.

16. The method of claim 15 wherein said normalized minutia is a skeletal minutia.

17. The method of claim 16 wherein said step of comparing includes a separate comparison of said angle identification templates against said minutia at a plurality of laterally displaced positions for each of said angle identification templates.

18. The machine method of enrolling a set of fingerprint minutia from a binary field of pixels having first and second binary signal values representing the fingerprint comprising the steps of:

generating an initial set of fingerprint minutia from an initial optical scan of a fingerprint to provide a field of minutia,

identifying the angle within said field of each of said minutia by comparing each of a set of angle identification minutia templates against each of said minutia,

repeating said steps of generating and identifying on repeated applications of the fingerprint that is subject to the initial scan, to provide a separate set of minutia on each of said subsequent applications of the fingerprint,

erecting a predetermined subfield zone centered upon each of said minutia from said initial scan,

correlating the position of said initial scan minutia against each of said subsequent scan minutia within the corresponding subfield zone,

designating as a minutia for the subsequent scan involved only those minutia whose position corresponds within predetermined limits with the position of the initial scan minutia, and

retaining in a final enrollment file only those minutia which are identified in a predetermined number of said scans to provide a stable, limited enrollment file.

19. The method of claim 18 wherein said step of correlating is limited to those minutia having an angular correspondence within predetermined limits.

20. The method of claim 19 wherein said step of identifying the angle of said minutia employs a normalized representation of each of said minutia in skeletal form.

21. The method of claim 20 wherein said step of repeating provides a first plurality of sets of minutia and a second plurality of sets of minutia, the correlation criteria between each minutia in each of said second plurality of sets and minutia from said initial scan being substantially tighter than the correlation criteria in each of said first plurality of sets of minutia, said second plurality of sets of minutia being generated and identified only for fingerprints having more than a predetermined number of minutia.

22. The machine method of enrolling a set of fingerprint minutia from a binary field of pixels having first and second binary signal values representing the fingerprint and of verifying correspondence to a binary field of pixels generated from a subsequent application of the fingerprint, comprising the steps of:

generating an initial set of fingerprint minutia from an initial optical scan of a fingerprint to provide a field of minutia,

repeatedly generating sets of fingerprint minutia from subsequent optical scans of repeated applications of the fingerprint that is subject to the initial scan,

retaining in a final enrollment file only those minutia which are identified in a predetermined number of said scan to provide a stable, limited enrollment file,

verifying the correspondence between minutia from a binary field of pixels representing a subsequent application of an input fingerprint and said minutia from said final enrollment file, said step of verifying employing analytical techniques and criteria substantially identical to those employed in said steps of repeatedly generating sets of minutia from repeated applications of the fingerprint during enrollment,

whereby the biases in verification are matched by and at least partially compensated by the comparable biases in the repeated scans and repeated steps of generating minutia during enrollment.

23. The method of claim 22 wherein said step of repeatedly generating includes a first set of subsequent scans and a second set of subsequent scans, the correlation criteria between each minutia in each of said second set of scans and minutia from said initial scan being substantially tighter than the correlation criteria in each of said first set of scans, said second set of scans applying only to individuals having more than a predetermined number of minutia.

24. The method of claim 23 wherein said step of correlating is limited to those minutia having an angular correspondence within predetermined limits.