Search

Finger vein, as an intrinsic trait, has received increasing attention from academia and industry for its liveness detection ability and high security. However, the principle of vein imaging and its contact-less mode results in low contrast of finger vein image, where especially the vein texture is susceptible to factors, such as scattering in skin and tissue, ambient illumination, and finger posture variation. To extract discriminative features from low contrast images, in this article, we propose an attention mechanism, namely, joint attention (JA) module, which enables dynamic adjustment and information aggregation in the spatial and channel dimensions of feature maps to focus on fine-grained details, thus enhancing the contribution of vein patterns to extract identity features. Besides, we embed a generalized mean (GeM) pooling layer into our network to reduce the dimensionality of feature maps and output a compact and highly expressive feature representation. Finally, we build a new finger vein authentication architecture, called JA finger vein network (JAFVNet), with competitive experimental results on multiple public and self-built datasets: 0.35% EER for SDUMLA-HMT, 0.08% EER for MMCBNU-6000, 0.34% EER for FV-USM, and 0.49% EER for FV-SCUT. Numerous ablation experiments further illustrate the effectiveness of each component in the network. [ABSTRACT FROM AUTHOR]

Despite claims that finger-vein biometrics can detect aliveness, recent research has shown that current systems can be fooled by forged vein patterns printed on a distinctive paper, raising considerable security concerns regarding the identification authenticity of these systems. Additionally, finger-vein identification exhibits low accuracy rates in real-world applications due to the inferior image quality caused by varied finger thicknesses and vein pattern variations caused by finger axial rotation. To address these issues, we propose a lightweight convolutional neural network (CNN) called the Finger-Vein Recognition and AntiSpoofing Network (FVRAS-Net), which integrates the recognition task and the antispoofing task into a unified CNN model by utilizing a multitask learning (MTL) approach and achieves high security and strong real-time performance. Then, a multi-intensity illumination strategy is introduced into the embedded biometric system to automatically select the most informative image for finger-vein identification, which can effectively improve the recognition performance of the real system. Finally, a challenging finger-vein database with images depicting severe axial finger rotation is built for more rigorous validation of the proposed system, which enriches the database resources for the finger-vein recognition community. Experiments demonstrate that the proposed FVRAS-Net achieves excellent performance in both recognition and antispoofing tasks on public data sets, especially on challenging databases with images depicting axial rotation. [ABSTRACT FROM AUTHOR]