Fusion2Print: Deep Flash-Non-Flash Fusion for Contactless Fingerprint Matching

TL;DR

Fusion2Print tackles the limited accuracy of contactless fingerprints by fusing paired flash and non-flash captures, enhancing ridge visibility through a dual-encoder fusion network and a color-space ridge enhancer, and learning cross-domain embeddings with TripletDistilNet. The paper introduces the Flash–Non-Flash Fingerphoto (FNF) Database to study modality Complementarity and demonstrates an end-to-end pipeline that achieves state-of-the-art cross-domain performance, with Embedding AUC up to $0.999$ and EER down to $1.12\%$. Key contributions include the FNF dataset, a fusion-enhancement-embedding framework, and comprehensive cross-domain verification results that outperform single-capture baselines. The work enables more accurate, hygienic, and interoperable fingerprint verification across contactless and contact-based systems, with potential for broader deployment and future improvements in robustness and multi-device evaluation.

Abstract

Contactless fingerprint recognition offers a hygienic and convenient alternative to contact-based systems, enabling rapid acquisition without latent prints, pressure artifacts, or hygiene risks. However, contactless images often show degraded ridge clarity due to illumination variation, subcutaneous skin discoloration, and specular reflections. Flash captures preserve ridge detail but introduce noise, whereas non-flash captures reduce noise but lower ridge contrast. We propose Fusion2Print (F2P), the first framework to systematically capture and fuse paired flash-non-flash contactless fingerprints. We construct a custom paired dataset, FNF Database, and perform manual flash-non-flash subtraction to isolate ridge-preserving signals. A lightweight attention-based fusion network also integrates both modalities, emphasizing informative channels and suppressing noise, and then a U-Net enhancement module produces an optimally weighted grayscale image. Finally, a deep embedding model with cross-domain compatibility, generates discriminative and robust representations in a unified embedding space compatible with both contactless and contact-based fingerprints for verification. F2P enhances ridge clarity and achieves superior recognition performance (AUC=0.999, EER=1.12%) over single-capture baselines (Verifinger, DeepPrint).

Figures (7)

• Figure 1: Overview of proposed fingerprint acquisition and enhancement framework: (a) The pipeline progressively enhances ridge–valley structure to produce an identity-discriminative embedding vector; (b) Embedding fine-tuning aligns domain-specific representations into a unified embedding space for contact and contactless fingerprints.
• Figure 2: Overview of the proposed Fusion2Print (F2P) framework. Aligned flash ($I_{\mathrm{flash}}$) and non-flash ($I$) images are processed by a dual-encoder attention fusion network, where $E_1$ and $E_2$ denote flash and non-flash encoders and $D$ produces the fused image $I_{\mathrm{fuse}}$. A U-Net--based enhancer refines $I_{\mathrm{fuse}}$ to obtain $I_{\mathrm{enh}}$, which is encoded by a ResNet-18-based feature extractor to produce discriminative embeddings. Verification is performed by comparing embeddings $e_1$ and $e_2$ from two capture sessions.
• Figure 3: Channel-wise RGB local contrast of $I_{\text{flash}}$, $I_{\text{diff}}$, and $I_{\text{fuse}}$.
• Figure 5: ROC and FRRvsFAR curves for the best TripletDistilNet model trained on $I_{\text{enh}}$.
• Figure 6: Activation maps of TripletDistilNet before and after fine-tuning, highlighting higher ridge- and core-focused attention (red) across contact and contactless domains.