Performances Enhancement of Fingerprint Recognition System Using Classifiers

Fingerprint recognition is best known and generally used as a biometric technology because of their high acceptability, immutability, and uniqueness. A fingerprint consists of ridges and valleys pattern also known as furrows. These patterns fully develop in the mother's womb and remain constant throughout the whole lifetime of the individual. The ridge bifurcation and ridge termination are the main minutiae features that are extracted for identification of individuals in fingerprint recognition system. The aim of this paper is to enhance the performance of the fingerprint recognition systems using classifiers. To achieve the aim, fingerprints from the FV2002 database are used, before these fingerprints are evaluated, image enhancement and binarization is applied as a pre-processing on fingerprints, by combining many methods to build a database of fingerprint features having minutia marking and minutia feature extraction. The fingerprint recognition is presented by image classification using MATLAB classifiers, i.e., Decision Tree, Linear Discriminant Analysis, medium Gaussian support vector machine (MG-SVM), fine K-nearest neighbor, and bagged tree ensemble. The aim of this paper is to make a comparison between classifiers for performance enhancement of the fingerprint recognition system. The MG-SVM classifiers significantly give the highest verification rate of 98.90% among all classifies used.