MegaMatcher On Card SDK

Smart card multi-biometrics

MegaMatcher On Card SDK offers matching-on-card technology that stores a person's fingerprint, iris and face templates on a smart card and performs template matching in a microprocessor embedded in the card, instead of matching biometric information on a PC processor.

The match-on-card method ensures that personal biometric information does not transfer to an external computer as it would in a more basic template-on-card system.

MegaMatcher On Card SDK is developed utilizing a set of ISO/IEC standards to enable interoperability with and easy integration into existing smart card and/or biometric systems.

Reliability and Performance Tests

The MegaMatcher On Card 10.0 template verification algorithm is a version of MegaMatcher algorithm adapted to the limited computational resources of smart cards. These tests were performed:

Reliability tests with publicly available databases for single biometric modalities

These reliability tests show the MegaMatcher On Card 10.0 fingerprint, face and iris single modality engines reliability. The tests were performed on the publicly available databases:

  • Fingerprint verification. The tests were performed using a subset of SONATEQ Fingerprint Database SQ FDB1-75TS1:
    • only left hand's index fingerprint images were used;
    • ISO/IEC 19794-2:2005 compact card minutiae format was used during testing;
    • the number of minutiae was truncated to 48 in both probe and gallery compact templates prior to matching;
    • ±45 degrees fingerprint rotation tolerance value was used for template matching in the maximized speed configuration;
    • ±180 degrees (unlimited) fingerprint rotation tolerance value was used for template matching in the maximized accuracy configuration.
  • Face verification. The tests were performed using face images from XM2VTS database. Proprietary compact template format was used during testing.
  • Iris verification. The tests were performed using iris images from ND-IRIS-0405 Iris Image Dataset:
    • proprietary compact template format was used during testing;
    • ± 15 degrees iris rotation tolerance value was used for template matching.

Receiver operation characteristic (ROC) curves are usually used to demonstrate the recognition quality of an algorithm. ROC curves show the dependence of false rejection rate (FRR) on the false acceptance rate (FAR).

Fingerprints
on SQ FDB1-75TS1
ROC chart: MegaMatcher On Card 10.0 fingerprint matching algorithm on SONATEQ database
Click to zoom
Faces
on XM2VTS
ROC chart: MegaMatcher On Card 10.0 face matching algorithm on XM2VTS database
Click to zoom
Irises
on ND-IRIS-0405
ROC chart: MegaMatcher On Card 10.0 iris matching algorithm on ND-IRIS-0405 database
Click to zoom
Fingerprints on SQ FDB1-75TS1
ROC chart: MegaMatcher On Card 10.0 fingerprint matching algorithm on SONATEQ database
Faces on XM2VTS
ROC chart: MegaMatcher On Card 10.0 face matching algorithm on XM2VTS database
Irises on ND-IRIS-0405
ROC chart: MegaMatcher On Card 10.0 iris matching algorithm on ND-IRIS-0405 database

Reliability tests with an internal Neurotechnology multi-biometric database

The tests with MegaMatcher On Card biometric fingerprint, face and iris matching engines and fused template matching algorithm were performed using a multi-biometric database:

  • The database had 7,500 sets of biometric records; each set contained 1 face, 2 irises and 10 fingerprints representing a unique person.
  • 1,500 unique persons were represented in the database.
  • 5 capture sessions were performed for each person.

The tests were performed with these biometric template types:

  • 1 fingerprint record – taken from left index fingerprint.
  • 1 face record.
  • 1 iris record – taken from left eye image.
  • 2 fingerprint records taken from same person's left and right index fingerprints.
  • 2 iris records taken from same person's different eyes.
  • 1 fingerprint + 1 face records left index fingerprint and face taken from the same person.
  • 1 face + 1 iris records left iris and face taken from the same person.
  • 1 fingerprint + 1 iris records left index fingerprint and left iris taken from the same person.
  • 1 fingerprint + 1 face + 1 iris records left index fingerprint, left iris and face taken from the same person.

The fingerprint template extraction and matching was performed with these settings:

  • the number of minutiae was truncated to 48 in both probe and gallery compact templates prior to matching;
  • ±45 degrees fingerprint rotation tolerance value was used for template matching in the maximized speed configuration;
  • ±180 degrees (unlimited) fingerprint rotation tolerance value was used for template matching in the maximized accuracy configuration.

These experiments were performed with the templates:

  • Experiment 1 maximized matching accuracy. The experiment was performed with all template types. MegaMatcher On Card 10.0 algorithm reliability in this test is shown as blue curves on the ROC charts.
  • Experiment 2 maximized matching speed. The experiment was performed only with template types that contained fingerprint records. MegaMatcher On Card 10.0 algorithm reliability in this test is shown as red curves on the ROC charts.

Receiver operation characteristic (ROC) curves are usually used to demonstrate the recognition quality of an algorithm. ROC curves show the dependence of false rejection rate (FRR) on the false acceptance rate (FAR).

1 fingerprint
ROC chart: MegaMatcher On Card 10.0 fingerprint matching algorithm
Click to zoom
1 face
ROC chart: MegaMatcher On Card 10.0 face matching algorithm
Click to zoom
1 iris
ROC chart: MegaMatcher On Card 10.0 iris matching algorithm
Click to zoom
     
2 fingerprints
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 2 fingerprints in a template
Click to zoom
2 irises
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 2 irises in a template
Click to zoom
Fingerprint + face
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 1 fingerprint and 1 face in a template
Click to zoom
     
Fingerprint + iris
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 1 fingerprint and 1 iris in a template
Click to zoom
Face + iris
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 1 face and 1 iris in a template
Click to zoom
Fingerprint + face + iris
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 1 fingerprint, 1 face and 1 iris in a template
Click to zoom
1 fingerprint
ROC chart: MegaMatcher On Card 10.0 fingerprint matching algorithm
1 face
ROC chart: MegaMatcher On Card 10.0 face matching algorithm
1 iris
ROC chart: MegaMatcher On Card 10.0 iris matching algorithm
2 fingerprints
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 2 fingerprints in a template
2 irises
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 2 irises in a template
Fingerprint + face
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 1 fingerprint and 1 face in a template
Fingerprint + iris
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 1 fingerprint and 1 iris in a template
Face + iris
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 1 face and 1 iris in a template
Fingerprint + face + iris
ROC chart: MegaMatcher On Card 10.0 fused matching algorithm, 1 fingerprint, 1 face and 1 iris in a template

Matching speed tests

MegaMatcher On Card 10.0 fingerprint, face and iris matching algorithms were tested on smart cards from several vendors. The matching speeds are available below. Please contact us to get more information about the expectations on a specific platform on which you intend to use it.

MegaMatcher On Card 10.0 genuine template verification time averages (seconds) (1)
Smart card model Fingerprints (2) Faces (3) Irises (3)
Java Card OS
(post-issuance application, maximized speed configuration)
≥ 0.666 ≥ 0.144 ≥ 0.385
Java Card OS
(native level, maximized accuracy configuration)
0.094 - 1.117 - -
Pure native OS
(native level, maximized accuracy configuration)
0.621 - 1.011 - -

(1) The performance on a specific smart card depends on the hardware chip platform, OS, also on-card matching algorithm integration level and the memory space it is being executed from. The verification times are provided as ranges, with the lowest values corresponding to the averages of the tests on appropriate OS smart cards with the fastest performance, and the highest values – with the slowest performance.
(2) Performance depends on the maximal number of minutiae features within enrolled and verified fingerprint templates. Results correspond to matching test of an enrolled and verified templates each containing 48 minutiae. Contact SCR335v1 USB smart card reader was used for PC/SC communication.
(3) Performance depends on the baud rate of contact or contactless protocol and APDU type chosen. Results correspond to matching face and iris compact card format templates using short length field APDUs. Timings are available for contact interface tests using SCR335v1 smart card reader.