Fingerprint reconstruction: from minutiae to phase

Area of Science:

• Biometrics
• Computer Vision
• Image Processing

Background:

• Minutiae-based fingerprint representation is widely used due to its compactness.
• Previous fingerprint reconstruction methods from minutiae templates often generate spurious minutiae or incomplete images.

Purpose of the Study:

• To develop a novel algorithm for reconstructing fingerprint images from minutiae templates.
• To evaluate the security implications of reconstructed fingerprints against recognition systems.

Main Methods:

• Representing fingerprint images as phase images, incorporating continuous and spiral phase components.
• Developing an algorithm to reconstruct the continuous phase from minutiae data.
• Converting the reconstructed phase image into a grayscale fingerprint image.

Main Results:

• The proposed algorithm successfully reconstructs whole fingerprint images.
• The reconstructed fingerprints contain significantly fewer spurious minutiae compared to previous methods.
• Evaluations demonstrated the feasibility of launching Type-I and Type-II attacks using reconstructed fingerprints.

Conclusions:

• The novel phase-based reconstruction algorithm effectively generates complete fingerprint images with high fidelity.
• Reconstructed fingerprints pose a security risk, as they can be successfully used in attacks against fingerprint recognition systems.