Fingerprint recognition with cellular partitioning and co-sinusoidal triplets

Abstract

In this fingerprint verification approach, a fingerprint image is divided into equally sized cells and the pattern is represented by a substitute resulting in a feature vector of fixed length. A related ISO standard recommends three different approaches for the selection of these. It suggests a cell size of approximately two ridges per cell. For the co-sinusoidal triplet approach for the retrieval of the spectral component this assumption was investigated. The influence of the cell size on the biometric performance was supported and additionally, a sound comparison method was implemented. To maintain a comprehensible evaluation the open Fingerprint Verification Competition (FVC) databases FVC2000 and FVC2002 were used.