Biometrics in New Zealand

Document Sample
Biometrics in New Zealand Powered By Docstoc
					Biometrics in New Zealand


  Passport issuing
  Border crossing
  System and information access
  Building access
                     ICAO STANDARDS

In May 2003 ICAO adopted a global harmonised blueprint for the
integration of biometric identification information into MRTD’s
 • Facial recognition was selected as the globally interoperable biometric for
   machine-assisted identity confirmation with MRTDs. Fingerprint and IRIS
   technology can also be used as an optional addition to facial recognition
 • High-capacity, contactless integrated circuit (IC) chips selected to store
   identification information in MRTDs
 • A specially developed logical data structure (LDS) will provide a framework
   for programming data to ensure interoperability of travel documents
 • A modified public key infrastructure (PKI) scheme will provide security of
   data stored in the IC chip against unauthorized alteration or access.
          U S Border Control Legislation
• Visa Waiver Requirements
   – NZ wishes to retain US Visa Waiver status
   – Work is underway to enable the issue of passports that incorporate
     biometric and document authentication identifiers from October
     2004
   – The new passport will incorporate Contact-less Chip technology to
     allow for the electronic storage of the holder’s portrait/s plus other
     data already displayed visually on the passport.
   – The new passport will enabe border control authorites to
     electronically check the documents authenticity and match the
     holder to the document using facial recognition technology
    New Zealand Passport Office



     Biometric Testing Programme

The use of Facial Recognition Technology in

     Passport Application Processing
       Biometrics Test Programme Outcome

To assess the suitability of facial recognition technology for use
    in NZ travel document issuance processes

Why Facial Recognition Technology?
   –    Minimal impact on the customer as Passports already collects a
        photo
   –    More socially acceptable in NZ
   –    Passports Database already holds an image of all current
        passport holders
   –    Aligns well with emerging international standards
                                Background

•       Testing programme commenced in February 2002 and is
        ongoing
    –     due to the rapid evolution of the technology (not backward
          compatible) and self imposed privacy constraints the programme
          has continued well beyond original timeframes
•       A test of the technology not simply of vendors products
•       Approximately 2.6 million images in the test database
•       The Programme is internationally recognised in the ICAO
        community and is being monitored by a number of US
        agencies (e.g DHS and the White House Technology
        Group)
          Specific Test Programme Objectives

•   Specifically, we are seeking to establish whether
    acceptable performance can be achieved in the following
    key areas
      •   The suitability of our current database of images for conversion to facial
          recognition templates
      •   Enrolment from a photograph supplied by the applicant, including the
          identification of best practice
      •   One to one matching (verification) of renewal photographs against
          existing image templates including the effects of aging
      •   One to many matching (identification) for images held under different
          identities (accuracy and search speed are the key factors)
      •   Image template storage and retrieval
            Overview of Results (to date)

•       Current database images are suitable for use with facial
        recognition technology
•       Photos submitted by applicants are suitable for use with
        facial recognition technology
•       One to one testing indicates that accuracy (probability of
        correct verification) up to ten years averaged 92.4%.
        Accuracy decreases with increasing elapsed time.
    –      Comparison of image templates captured within the same year
           produced accuracy levels of 97% however in the tenth year
           accuracy dropped to 88%
        Further Results (to date)

•   Early results of one to many testing indicates that
    we can accurately identify a person posing as
    another 60% of the time when compared with 2.6
    million others
•   Adding gender and age-band filters increases
    performance to around 78%
•   One to many searches can be carried out inside
    four seconds on a database of 2.6 million
         Further Development & Testing

• Filtering to improve accuracy of fraud detection
• 3D modeling of the facial image
• Image capture
   – optical image vs digital scanning
• Test in operational mode
• Report findings to international ICAO community
• Assess suitability for implementation
Questions
   and
Discussion