An Investigation of Benford’s Law Divergence and Machine Learning Techniques for Intra-Class Separability of Fingerprint Images

Abstract

Protecting a biometric fingerprint database against attackers is very vital in order to protect against false acceptance rate or false rejection rate. A key property in distinguishing biometric fingerprint images is by exploiting the characteristics of these different types of fingerprint images. The aim of this paper is to perform an intra-class classification of fingerprint images using Benford's law divergence values and machine learning techniques. The usage of these Benford’s law divergence values as features fed into the machine learning techniques has proved to be very effective and efficient in the intra-class classification of biometric fingerprint images. The effectiveness of our proposed methodology was demonstrated on five datasets resulting in a total of 367 samples. All the machine learning techniques used in this experiment were trained using the k-fold cross validation and the dataset was split into ten times (10-folds). The models achieved high intra-class classification mean accuracies of 99.72% for the Convolutional Neural Networks (CNN), and 95.90% for the Naïve Bayes. Again, the Decision Tree and Logistic Regression, achieved accuracies of 95.62%, and 94.47%, respectively. These results showed that Benford’s law features and machine learning techniques, especially the CNN and Naïve Bayes can be effectively applied for the intra-class classification of fingerprint images. The implication of these results is that the different types of fingerprint images can be effectively discriminated using Benford's law divergence values and machine learning technique for forensics and biometrics applications.