3D CNN-based fingerprint anti-spoofing through optical coherence tomography

Area of Science:

• Biometrics
• Medical Imaging
• Artificial Intelligence

Background:

• Optical coherence tomography (OCT) provides noninvasive, high-resolution imaging of subsurface tissue structures.
• Fingerprint recognition systems are vulnerable to spoofing attacks using artificial fingerprints.
• Internal fingerprint structures, including sweat glands, offer unique biometric features for enhanced security.

Purpose of the Study:

• To propose and evaluate a novel anti-spoofing method for fingerprint recognition using OCT.
• To leverage the 3D spatial continuity of biometric data from OCT scans.
• To demonstrate the effectiveness of a 3D convolutional neural network (CNN) for OCT-based fingerprint anti-spoofing.

Main Methods:

• Acquisition of internal fingerprint structures using Optical Coherence Tomography (OCT).
• Development of a 3D Convolutional Neural Network (CNN) model to analyze OCT data.
• Training and testing the 3D CNN model on self-built and public fingerprint datasets.
• Comparison of the 3D CNN method against classic network architectures.

Main Results:

• The proposed OCT fingerprint anti-spoofing method demonstrated high accuracy.
• The 3D CNN approach effectively utilized the spatial continuity of 3D biometric data.
• The 3D CNN strategy outperformed traditional network models in anti-spoofing experiments.

Conclusions:

• OCT technology is a viable tool for acquiring detailed internal fingerprint structures for security applications.
• 3D CNNs are effective for analyzing OCT data and enhancing fingerprint anti-spoofing capabilities.
• The proposed method offers a promising solution to improve the security and reliability of biometric systems.