home  |   research   |  members   |  projects  |  publications   |  conferences of interest   |  downloads   |  contact  |  intern
 
 

Finger Vein Recognition and Intra-Subject Similarity Evaluation of Finger Veins using the CNN Triplet Loss

This is "The Multimedia Signal Processing and Security Lab", short WaveLab, website. We are a research group at the Computer Sciences Department of the University of Salzburg headed by Andreas Uhl. The short name "WaveLab" already indicates that wavelets are among our favorite tools - we have 15 years of experience in this area. Our research is focused on Multimedia Security including Watermarking, Image and Video Compression, Medical Image Classification, and Biometrics.

Abstract

Finger vein recognition deals with the identificationof subjects based on their venous pattern within the fingers.There is a lot of prior work using hand crafted features, butonly little work using CNN based recognition systems. Thisarticle proposes a new approach using CNNs that utilizes thetriplet loss function together with hard triplet online selection forfinger vein recognition. The CNNs are used for three differentuse cases: (1) the classical recognition use case, where everyfinger of a subject is considered as a separate class, (2) anevaluation of the similarity of left and right hand fingers fromthe same subject and (3) an evaluation of the similarity ofdifferent fingers of the same subject. The results show thatthe proposed nets achieve superior results compared to priorwork on finger vein recognition using the triplet loss function.Furtherly, we show that different fingers of the same subject,especially symmetric fingers (same finger type but from differenthand), show enough similarities to perform recognition. The laststatement contradicts the current understanding in the literaturefor finger vein biometry, in which it is assumed that differentfingers of the same subject are unique identities.

Reference

[Wimmer20a   ] Finger Vein Recognition and Intra-Subject Similarity Evaluation of Finger Veins using the CNN Triplet Loss Georg Wimmer, Bernhard Prommegger, Andreas Uhl In Proceedings of the 25th International Conference on Pattern Recognition (ICPR), pp. 1-7, 2020, accepted

Data Sets

The Hong Kong Polytechnic University Finger Image Database

The Hong Kong Polytechnic University Finger Image Database (HKPU) Database consists of simultaneously acquired finger vein and finger surface texture images. It contains finger vein images from 156 volunteers, four fingers each (left and right index and middle finger). The data was captured in two different sessions, capturing six samples per finger in each session. Further information regarding the data set can be found by the following link:

HKPU Download Page (external link)

PLUSVein-FV3 Finger Vein Data Set

The PLUSVein-FV3 Finger Vein Data Set (PLUSVein-FV3) is a publically available fingr vein data set. It contains palmar and dorsal images of 360 fingers from 60 different subjects (ring, middle and index finger from both hands) captured in one session with five samples per finger using two different variants of the same sensor: One utilizing NIR laser modules for illumination, the other one using NIR LEDs. Further information regarding the data set can be found by the following link:

PLUSVein-FV3 Data Set

SDUMLA-HMT Database

The SDUMLA-HMT database is a multimodal biometric database that contains samples for face, gait, iris, fingerprint and finger veins from 106 individuals. The finger vein subset contains six fingers (ring, middle and index finger from both hands) per subject, captured in one session taking six images of each finger. Further information regarding the data set can be found by the following link:

SDUMLA-HMT Download Page (external link)

University of Twente Finger Vascular Pattern Database

The University of Twente Finger Vascular Pattern (UTFVP) Database is a publically available finger vein data set. It contains six fingers (ring, middle and index finger from both hands) from 60 volunteers acquired in two sessions. Further information regarding the data set can be found by the following link:

UTFVP Download Page (external link)