A Novel 2D Contactless Fingerprint Matching Method

Lei Shi,Sheng Lan,Hao Gui,Yujiu Yang,Zhenhua Guo
DOI: https://doi.org/10.1016/j.neucom.2022.05.092
IF: 6
2022-01-01
Neurocomputing
Abstract:Fingerprint recognition is an important biometric technology. According to the data collection methods, fingerprints can be divided into contact-based and contactless schemes. In recent years, 2D contactless fingerprints have constantly received attention and research attention because of their safety and hygiene advantages and low cost. However, because of its collection mode, contactless fingerprints usually have low-resolution and are prone to posture inconsistency. This issue leads to the poor matching effect of most current fingerprint algorithms on 2D contactless fingerprints, especially for mainstream minutiae-based methods. To address these problems, we propose a novel 2D contactless fingerprint matching method named as fingerprint triplet-GAN (FTG). It is an end-to-end method consisting of a triplet network and a GAN, and it bypasses the conventional extraction of fingerprint minutiae. First, we use a triplet network to perform data balancing and augmentation, and improve the robustness at low-resolution. Furthermore, we design a GAN to help extract posture-independent features to improve the rotation robustness of the model. The effectiveness of the proposed method is evaluated on two databases: PolyU and FVC2004DB1. The experimental results show that it can achieve state-of-the-art performance on 2D contactless fingerprints. Codes are available at: https://github.com/IC-LAB.
What problem does this paper attempt to address?