.png){kind=icon}
.png){kind=icon}
Return
Vein Pattern-Based Partial Finger Vein Alignment and Recognition
.png){kind=icon}
PRE
AI
DOI：10.1109/TBIOM.2025.3592306.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Partial finger vein recognition is a challenging but important task in scenarios where the sensors used for user enrollment and recognition differ due to sensor upgrades, and there is a significant disparity in the imaging area between their respective imaging windows. Although the state-of-the-art recognition methods achieve promising performance on full finger vein images, they may suffer from degradation on partial finger vein images. To deal with the problem of recognizing a patch of a finger vein image, this paper proposes a vein pattern-based partial finger vein alignment and recognition method. This method employs the direction variation points as minutiae of finger vein pattern in conjunction with the vein bifurcation points and endpoints to align full and partial images. The process involves a two-stage alignment mechanism, i.e., rough alignment constrained by finger physical structure, and precise alignment determined by joint texture and location features. The candidate matching region(s) can be identified within the full gallery image corresponding to the partial probe image, and further used in subsequent minutiae and vein pattern-based recognition. Gallery images that fail to exhibit minutiae matches are classified as imposters in verification mode, or receive matching scores of zero in identification mode. The extensive experimental results on three finger vein databases demonstrate the advantage of the proposed method in partial finger vein recognition, achieving an accuracies of 97.54% on HKPU-FV, 97.22% on PLUS-LED and 97.22% on PLUS-LAS.
Keywords:
Fingers
Veins
Image recognition
Pattern recognition
Pattern matching
Probes
Feature extraction
Biometrics
Databases
Convolutional neural networks
Partial finger vein recognition
vein pattern
image alignment
finger physical structure
Journal
I
IF:
0
Papers:
67
Citations:
0