Fingerprint Classification with reduced penetration rate : Using Convolutional Neural Network and DeepLearning

Abstract

Biometric fingerprint feature extraction is a complex process, but it can be simplified by using many applications. When dealing with poor quality fingerprint images the performance of the traditional minutiae detection algorithm found to be deteriorated. Recognizing the fingerprint in a fast and flexible method is hot research topics in these days because of the matured fingerprint identification technology and the massive fingerprint database. This paper mostly focused on fingerprint classification using convolutional neural networks which excludes the need of specific feature extraction process. In such situation, this method proved to predict a class even with poor quality fingerprint image that are commonly rejected by most of the algorithm. This study also gone through to minimize the penetration rate in the database. Various plots of our work shows good accuracy rate and better penetration rate.