作者: A.K. Jain , S. Minut
DOI: 10.1109/ICPR.2002.1048340
关键词: Fingerprint 、 Flow field 、 Whorl (botany) 、 NIST 、 Minutiae 、 Computation 、 Impression 、 Pattern recognition 、 Mathematics 、 Computer vision 、 Contextual image classification 、 Artificial intelligence
摘要: Fingerprint classification consists of labeling a fingerprint impression as one several major types fingerprints: arch, left loop, right whorl, etc. The problem matching amounts to deciding whether or not two impressions were produced by the same finger. We propose model based method for which only uses flow field, avoiding non-trivial computation thinned ridges and minutia points. For each class, kernel is defined, models shape fingerprints in that class. then achieved finding best fits field given fingerprint. obtain accuracy 91.25% on NIST 4 database. also show how fitting procedure can be used alignment.