Illumination-Aware Contactless Fingerprint Spoof Detection via Paired Flash-Non-Flash Imaging

Roja Sahoo; Anoop Namboodiri

cs.CVMar 18, 2026

Illumination-Aware Contactless Fingerprint Spoof Detection via Paired Flash-Non-Flash Imaging

Authors: Roja Sahoo, Anoop Namboodiri

Organizations: IIIT Hyderabad, Gachibowli, India

Abstract

Contactless fingerprint recognition enables hygienic and convenient biometric authentication but poses new challenges for spoof detection due to the absence of physical contact and traditional liveness cues. Most existing methods rely on single-image acquisition and appearance-based features, which often generalize poorly across devices, capture conditions, and spoof materials. In this work, we study paired flash-non-flash contactless fingerprint acquisition as a lightweight active sensing mechanism for spoof detection. Through a preliminary empirical analysis, we show that flash illumination accentuates material- and structure-dependent properties, including ridge visibility, subsurface scattering, micro-geometry, and surface oils, while non-flash images provide a baseline appearance context. We analyze lighting-induced differences using interpretable metrics such as inter-channel correlation, specular reflection characteristics, texture realism, and differential imaging. These complementary features help discriminate genuine fingerprints from printed, digital, and molded presentation attacks. We further examine the limitations of paired acquisition, including sensitivity to imaging settings, dataset scale, and emerging high-fidelity spoofs. Our findings demonstrate the potential of illumination-aware analysis to improve robustness and interpretability in contactless fingerprint presentation attack detection, motivating future work on paired acquisition and physics-informed feature design. Code is available in the repository.

Figures & tables

Fig. 1: Blockwise OCL maps for sample flash and non-flash fingerprint images. Brighter regions(higher intensity) indicate higher ridge clarity; flash images show consistently stronger ridge coherence.

Fig. 2: LCS and ridge-valley intensity profiles for sample flash and non-flash fingerprint blocks, showing improved ridge-valley contrast under flash illumination.

Fig. 3: Comparison of standard (AIT Sharpness, NFIQ2) and custom patch-wise image quality metrics over the sample dataset, where flash images outperform non-flash images across all measures.

Fig. 4: Photometric analysis of local contrast and edge energy for R, G, and B channels of the sample dataset under flash and non-flash illumination.

Panel of Figure 5 — Fig. 5: Attention maps for sample flash and non-flash contactless fingerprint images. (a)–(b) Baseline DINOv2 attention maps with limited ridge awareness but improved spatial structure under flash lighting. (c)–(d) Fine-tuned ResNet-18–based model showing enhanced ridge-focused attention, especially for flash images.

Model	G-Non-Flash	G-Flash	S-Non-Flash	S-Flash
Baseline	$0.21\pm 0.19$	$\mathbf{0.23\pm 0.16}$	$0.22\pm 0.24$	$0.23\pm 0.21$
Fine-tuned	$0.37\pm 0.16$	$\mathbf{0.49\pm 0.13}$	$0.26\pm 0.22$	$0.27\pm 0.14$

TABLE I: Activation statistics for genuine (G) and spoof (S) fingerprints under flash and non-flash conditions for the ResNet-18–based model, computed over the sample dataset (mean

\pm

std).

PAI Type	Typical Forms	Material / Method Weakness	Cues Exposed by Flash–Non-Flash
Printed spoofs	Paper printouts (inkjet, laser)	Flat surface, no subsurface scattering; ink-dependent color response; uniform reflectance	Exaggerated specular highlights under flash; ink-induced RGB distortions; low micro-texture consistency across illumination
Digital spoofs	Smartphone / tablet displays	Emissive surface; pixel grid structure; channel saturation	Screen glare and moiré patterns under flash; unnatural channel peaks; inconsistent photometric response
Physical / molded spoofs	Silicone, latex, gelatin, resin replicas; prosthetic fingers (closely matching human skin tone)	Surface-only reflection; limited or no subsurface scattering; uniform ridge depth	Uniform specular reflection under flash; reduced ridge-valley micro-variation; inconsistent differential response
3D modeled spoofs	3D replicas generated from 2D contactless images	Synthetic geometry; missing organic ridge irregularities; material homogeneity	Illumination-invariant texture patterns; weak flash–non-flash differences; absence of oil- or sweat-induced highlights

TABLE II: Summary of contactless fingerprint PAIs and illumination-dependent cues exposed by paired flash–non-flash acquisition.

Panel of Figure 8 — Fig. 6: Inter-channel correlation analysis for the sample dataset of flash and non-flash fingerprint images and their spoof counterparts. (a) Pearson correlation and (b) mutual information heatmaps show stronger off-diagonal channel relationships for genuine images, with clearer distinction from spoof under flash illumination. Mann–Whitney U tests confirm the statistical significance of off-diagonal activations (Pearson: $p<0.001$ for both conditions; MI: $p<0.001$ for flash and $p=0.035$ for non-flash).

Sample Type	SHR
Real Fingerprints	0.009
Spoof Fingerprints	0.043

TABLE III: Average Specular Highlight Ratio (SHR) over the dataset of real and spoof fingerprints using paired flash and non-flash images. Higher SHR indicates increased specular reflection and is more indicative of spoof fingerprints.

Panel of Figure 10 — Fig. 7: Log-magnitude FFTs and radially averaged spectra are shown for (a) flash and (b) non-flash contactless fingerprint images of genuine and spoof samples. Texture realism ratio (fraction of blocks with dominant FFT peaks) of the dataset is higher for spoof fingerprints (0.226) than real fingerprints (0.087), indicating disrupted, grid-like textures and increased high-frequency artifacts enabling a clear distinction.

Explore similar work

Generation of Synthetic Fingerphotos with GANs
Aug 15, 2026Conor Miller-Lynch, Sandip Purnapatra, Syed Konain Abbas +4Synthetic Data Generation Synthetic Data Evaluation
UNC Charlotte, Charlotte, NC, USA · Clarkson University, Potsdam, NY, USA
Identity-Consistent Multi-Pose Generation of Contactless Fingerprints
May 5, 2026Zhiyu Pan, Xiongjun Guan, Jianjiang Feng +1Synthetic Data Augmentation Biometric Identification
Department of Automation, Tsinghua University, Beijing 100084, China
CAAP: Capture-Aware Adversarial Patch Attacks on Palmprint Recognition Models
Apr 8, 2026Renyang Liu, Jiale Li, Jie Zhang +6Adversarial Patch Attacks Physical Adversarial Attacks
Institute of Data Science, National University of Singapore, Singapore 117602, Singapore · Centre for Frontier AI Research (CFAR), A*STAR, Singapore, 138634 · School of Cyber Science and Engineering, Wuhan University, China, 430072 +2

Aug 15, 2026J/K move · Enter open · S save

Generation of Synthetic Fingerphotos with GANs

Conor Miller-Lynch, Sandip Purnapatra, Syed Konain Abbas, Lambert Igene+3

UNC Charlotte, Charlotte, NC, USA · Clarkson University, Potsdam, NY, USA

Synthetic Data Generation Synthetic Data Evaluation Biometric Identification Generative Adversarial Networks

Open paper AskPDF arXiv

Contactless fingerprinting is an emerging approach to biometric authentication that allows users to scan their fingerprints without touching a scanner. Due to the limited amount of contactless fingerprint data available and the security risks associated with sharing real individuals' fingerprints, it is valuable to explore methods of generating synthetic data that can be used in place of - or in conjunction with - real data to develop and evaluate contactless fingerprinting systems. In this paper, we present and evaluate synthetic fingerphotos generated using StyleGAN2-ADA and StyleGAN3, existing image generation architectures. We evaluate the realism, privacy preservation, and variety of the synthetic fingerphotos by comparing their biometric feature statistics to those of real fingerphotos, computing match scores between real and synthetic fingerphotos, and computing match scores between different synthetic fingerphotos. This paper provides a quantitative comparison point for future evaluations of synthetic fingerphotos. The evaluation code is made available at https://github.com/cmillerlynch/fingerphoto-gan.

Abstract

Explore similar work

Generation of Synthetic Fingerphotos with GANs

Identity-Consistent Multi-Pose Generation of Contactless Fingerprints

CAAP: Capture-Aware Adversarial Patch Attacks on Palmprint Recognition Models