이웃 특징점 부스트 기반 새로운 지문 매처

초록

Most fingerprint matching algorithms perform the matching based on a fixed tolerance box that is not flexible enough. This paper proposes a neighbor minutiae?boost method with very few soft thresholds to improve the matching of fingerprints. First, an intra-probability table is computed in which each node represents the matching probability of a minutia pair boosted by neighbor minutiae. Then, minutia pairs with high probabilities are selected as the candidate matched minutia pairs. Finally, these minutia pairs are verified by counting the ridges and the orientation field difference. In addition, the missing minutiae in the common area can be found via retrieval steps after verifying matched minutia pairs. Preliminary experimental results show that the proposed method outperforms Bozorth3 in terms of matching performance