An Efficient Machine Learning Approach for Fingerprint Authentication Using Artificial Neural Networks

Fingerprint identification is a biometric technique that identifies persons by using categorization models and computations on images at the most basic level. For feature extraction, a fingerprint detection framework that employs a number of image pre-processing methods as well as a variety of image locality bifurcations is described and assessed. For classification, an Artificial Neural Network (ANN) is suggested. This paper introduces the Gabor filterfor feature extraction and the ANN machine learning method for classification. The feature vector is created by combining Gabor filtering features with machine learning techniques such as Artificial Neural Networks. Using the extracted features, a multiclass classifier was developed. The experiments in this study were conducted using the NIST Special Database 4 (NISTSD4). In addition, the confusion matrix supported these findings and the proposed approach in terms of accuracy (97.95 %) outperformed more recent Machine learning classification techniques such as Random Forest, KNN, Support Vector Machine and Decision Tree.