Fingerprint classification based on a Q-Gaussian multiclass support vector machine

Accurate recognition and actual classification of fingerprint are vital and necessary for fingerprint identification. Previous researchers have used many classification algorithms to develop fingerprint classification model, but they still have some certain problems like time of implementation to do the task, cost of implementation, working on non-linear features, working on multi-dimensional features and under or over learning problems. In this paper, a Q-Gaussian multi-class support vector machine (QG-MSVM) for fingerprint classification is proposed in which Q-Gaussian function is incorporated into SVM as a kernel function. The proposed method is tested in CASIA, FVC2000, FVC2002 and FVC2004 databases and compared with the MSVM methods with linear kernel, Gaussian Radial Basis Function kernel (RBF), Polynomial kernel and other state-of-the-art methods. The experimental results show that QG-MSVM demonstrates better performance than other classifiers and overcome many MSVM problems. The overall performance of the QG-MSVM classifier is comprehensively superior to all others.