A Novel Fingerprint Recogntion Using Minutiae Features

Fingerprint Recognition has been widely used for personal authentication for several centuries. Most of these systems are based on minutiae features like ridge bifurcation and ridge ending. In this paper we present fingerprint recognition by using only ridge ending as the feature vector. A new model is proposed for both denoising and segmentation and the minutiae points are extracted directly from grayscale images without binarization ad thinning. Recognition is performed using classification ensembling with BPNN, SVM and KNN as classifiers. Experimental results obtained using FVC2004 database show that the proposed method works efficiently in a faster and improved manner.