ColFigPhotoAttnNet: Reliable Finger Photo Presentation Attack Detection Leveraging Window-Attention on Color Spaces

2025-03-07

AI Summary

To address poor cross-device generalization and weak robustness in smartphone fingerprint photo presentation attack detection (PAD), this paper proposes a novel architecture integrating color-space windowed attention with a nested residual network. First, it systematically analyzes the performance degradation mechanisms of CNN/Transformer models across heterogeneous acquisition devices. Second, it introduces a channel-level windowed attention mechanism—novel in PAD—to enhance adaptability to hardware diversity. Third, it jointly models multi-color-space features and employs a deep nested residual structure to improve fine-grained texture discrimination. Evaluated on a cross-device multi-database comprising fingerprints captured from iPhone 13 Pro, Pixel 3, Nokia C5, and OnePlus One, the method achieves an average ACER reduction of 37.2% across three mainstream PAD benchmarks, significantly outperforming state-of-the-art approaches and demonstrating superior generalization and reliability.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search

Abstract

Finger photo Presentation Attack Detection (PAD) can significantly strengthen smartphone device security. However, these algorithms are trained to detect certain types of attacks. Furthermore, they are designed to operate on images acquired by specific capture devices, leading to poor generalization and a lack of robustness in handling the evolving nature of mobile hardware. The proposed investigation is the first to systematically analyze the performance degradation of existing deep learning PAD systems, convolutional and transformers, in cross-capture device settings. In this paper, we introduce the ColFigPhotoAttnNet architecture designed based on window attention on color channels, followed by the nested residual network as the predictor to achieve a reliable PAD. Extensive experiments using various capture devices, including iPhone13 Pro, GooglePixel 3, Nokia C5, and OnePlusOne, were carried out to evaluate the performance of proposed and existing methods on three publicly available databases. The findings underscore the effectiveness of our approach.

Problem

• Detects finger photo presentation attacks on smartphones
• Addresses poor generalization across different capture devices
• Improves robustness against evolving mobile hardware

Innovation

• Window-attention on color channels for PAD
• Nested residual network as predictor
• Cross-device performance evaluation