BPFNet: A Unified Framework for Bimodal Palmprint Alignment and Fusion

Bimodal palmprint recognition use palm vein and images at the same time, which can achieve high accuracy has intrinsic anti-falsification property. For bimodal verification, ROI detection alignment of region-of-interest (ROI) are two crucial points for matching. Most existing plamprint methods based on keypoint algorithms, however difficulties lying in tasks make results not accurate. Besides, these feature fusion algorithms image-level fully investigated. To improve performance bridge gap, we propose our Palmprint Fusion Network (BPFNet) focuses localization, image fusion. BPFNet is an end-to-end deep learning framework contains parts: The network directly regresses ROIs conducts by estimating translation. In downstream, leveraging a novel cross-modal selection scheme. demonstrate effectiveness BPFNet, implement experiments touchless datasets proposed achieves state-of-the-art performances.