Iris Grid Image Classification using Naive Bayes for Human Biometric System

Abstract

Biometrics is a measurement of a person's physical and behavioral characteristics. Iris image is one of many biometrics data such as fingerprint, voice, face, and gait that can be used as an identifier. Iris is the colored part of the eye that helps the pupil see clearly and regulates light entry. Iris recognition is one of the important topics in biometric systems because of its unique pattern. Several related studies have been carried out to automatically obtain the most efficient method to understand and recognize the iris for human verification. This study proposes an analysis of iris images for biometrics systems with effective image processing techniques for system recognition. CVBL Iris image dataset was used in this study with 4320 iris images. After converting the iris image into a rectangle form, the Grid iris image experiment was implemented to find the highest accuracy. Several iris image grid-size were simulated to find the best accuracy. Multinomial Naive Bayes is used as a classifier. The Naive Bayes method is a machine learning method that uses probability calculations (rules-based). This algorithm uses probability and statistical methods, which predict future probabilities based on the previous data. The study results indicate that the proposed method can recognize the iris by identifying its fibers and encoding the fibers data using a grid image approach, with a classification accuracy of 92.37%, using an iris grid size of 70x50 pixels. This research can be useful for developing human biometric systems based on iris with a simple preprocessing approach.