Contactless Fingerprint Enhancement Algorithm Based on Frequency Domain and Multi-Channel Gabor Filtering

Fingerprints are among the most widely used biometric identifiers due to their exceptional uniqueness, accuracy, and durability. Contactless fingerprints offer significant advantages over contact fingerprints regarding hygiene and security, making them an excellent alternative for completing biometric identification. However, the imaging quality of contactless fingerprints is subpar, particularly due to the contrast between the ridges and valleys, which can easily result in the generation of false shadows. However, its ridge-valley structure is relatively continuous, and due to perspective distortion, it is easy to produce ridges with uneven frequency distribution. To address the above problems, we propose an enhancement scheme for contactless fingerprints. The fingerprint orientation map and frequency map are obtained through frequency domain analysis. The estimation of this frequency map does not depend on the accuracy of the orientation map, reduces the coupling between attribute maps, and introduces multi-channel filtering to adapt to the characteristics of contactless fingerprints. We verified the effectiveness of the proposed method through simulation experiments. We completed the verification experiment using NIST on two public contactless fingerprint data sets, and compared it with the traditional Gabor and Fourier transform analysis and reconstruction algorithms, achieving a lower equal error rate. The effectiveness of the method was verified through ROC curve visualization, proving that the method is an effective solution for contactless fingerprint enhancement.