Towards Fingerprint Mosaicking Artifact Detection: A Self-Supervised Deep Learning Approach

TL;DR

The paper tackles hard mosaicking artifacts in fingerprint images by introducing a self-supervised deep learning detector built on a UNet++ architecture with a ResNeSt-50d encoder. It learns robust representations from unlabeled data through synthetic patch- and line-based artifact injections and outputs both segmentation masks and a novel mosaicking artifact score $S$, defined with patch weight $b$ and line weight $c$. The authors validate across multiple modalities including contactless, rolled, and pressed fingerprints, and analyze the impact of artifacts on equal-error-rate (EER) through several ABIS, demonstrating significant performance gains in artifact detection and potential improvements in biometric data quality control. The approach provides cross-domain robustness, a scalable evaluation metric, and practical implications for sensor design and biometric security implementations.

Abstract

Fingerprint mosaicking, which is the process of combining multiple fingerprint images into a single master fingerprint, is an essential process in modern biometric systems. However, it is prone to errors that can significantly degrade fingerprint image quality. This paper proposes a novel deep learning-based approach to detect and score mosaicking artifacts in fingerprint images. Our method leverages a self-supervised learning framework to train a model on large-scale unlabeled fingerprint data, eliminating the need for manual artifact annotation. The proposed model effectively identifies mosaicking errors, achieving high accuracy on various fingerprint modalities, including contactless, rolled, and pressed fingerprints and furthermore proves to be robust to different data sources. Additionally, we introduce a novel mosaicking artifact score to quantify the severity of errors, enabling automated evaluation of fingerprint images. By addressing the challenges of mosaicking artifact detection, our work contributes to improving the accuracy and reliability of fingerprint-based biometric systems.

Figures (4)

• Figure 1: Depiction of different artifact types created for supervisory signal
• Figure 2: Model architecture of our proposed combination of ResNeSt for the encoder and UNet++ for the general model architecture and decoder design.
• Figure 3: Exemplary results with blurred input images for privacy protection.
• Figure 4: Synthetic fingerprint alterations based on SynColFinGe applied to SFinGe generated fingerprints.