Fingerprint recognition using model-based density map.

Area of Science:

• Biometrics
• Computer Vision
• Pattern Recognition

Background:

• Traditional fingerprint recognition relies heavily on minutiae points.
• Large-scale applications require more discriminative features for accurate identification.
• Minutiae-based methods can be insufficient for handling large databases and variations in fingerprint quality.

Purpose of the Study:

• To propose a novel fingerprint feature representation using a polynomial model of fingerprint density maps.
• To integrate density map features with conventional minutiae-based matching for improved recognition performance.
• To reduce the additional storage cost while enhancing the accuracy of fingerprint recognition systems.

Main Methods:

• Developed a polynomial model to approximate the density map of fingerprints.
• Extracted model parameters as novel features for fingerprint representation.
• Implemented a decision-level fusion scheme to combine density map matching with minutiae-based matching.

Main Results:

• The proposed polynomial model effectively captures fingerprint density information.
• Integrating density features with minutiae matching resulted in significantly better performance compared to minutiae-based matching alone.
• The approach offers a low additional storage cost for enhanced feature representation.

Conclusions:

• Fingerprint density map information, modeled polynomially, provides valuable features for recognition.
• Decision-level fusion of density and minutiae features enhances large-scale fingerprint recognition accuracy.
• This method offers a practical and efficient solution for improving biometric identification systems.