Development of a Naive Bayes classifier for image quality assessment in biometric face images

Resumo

Currently, the use of applications that requiring biometric facial authentication is increasing. However, the performance of these systems is compromised when the images do not present the quality required for their operation. Considering this, in the present work, a Naive Bayes classifier was developed for image quality assessment (IQA) dedicated to facial biometry systems. The metrics that characterize the image according to blur, sharpness and focus are used as input variables. A bibliographical study is carried out on the subject, deepening in the characteristics of the Bayesian classifiers. A public database is used to create the statistical data with which the Naive Bayes classifier is developed. Finally, the results of the classifier are presented for a set of conditions of the predictor variables.