Fusion of presentation attack detection (liveness) and match score database

This database consists of match and presentation attack detection (liveness) scores generated from 8019 total live images and 2525 total spoof images. The fingerprint images were captured from the right thumb (R1) and ridge index finger (R2) from each subject. The spoof images are collected from fake fingers, created from molds of the subjects in the live set. The total number of subjects in the database is 49, split into two subsets for training and testing purposes. The training set consists of 25 of the subjects and the test set consists of 24 of the subjects. Three groups of scores are provided for each subset; Normal Presentation—Genuine, Normal Presentation—Imposter, and Presentation Attack (from genuine users), totaling six sets of scores. Each presentation includes three scores; one match score comparing gallery and probe images and two liveness scores generated from probe image. The first liveness score is determined by a simple liveness algorithm, and the second liveness score is determined by a more advanced liveness algorithm. The number of presentations included in each set of scores is shown in the table below.

The purpose of this database is for the development and/or evaluation of biometric fusion methods for incorporating a liveness detection module into a fingerprint authentication system.

Subset
# Subj
# live
images
# spoof
images
Normal Presentation-
Geniune
Normal Presentation-
Imposter
Attack Presentation-
(Geniune)
Train 25

R1: 2,187

R2: 1,896

R1: 724

R2: 491

519,198 911,476 106,943
Test 25

R1: 2,153

R2: 1,783

R1: 749

R2: 561

381,182 976,161 132,075

When using biometric collection, please cite:

Peter Johnson, Stephanie Schuckers, Evaluation of Presentation Attack Detection:  An Example, IBPC 2014.