Performance Evaluation of Feature Sets of Minutiae Quadruplets

Abstract

The features proposed in this paper are derived from minutiae quadruplets and are applicable in matching and indexing ngerprint images. In this work nineteen different possibilities of features were explored for indexing and the performances of some of the feature sets were mixed: some giving good performances on certain databases and poor performances on other databases. A nal ranking was done and one feature-set was chosen as viable geometrical features for minutiae matching and indexing based on their performances on three Fingerprint Verication Databases (FVC) 2000, 2002 and 2004.