A Novel Iris Based Security Policy Using Ldp And Ensemble Classification
N. Susitha, Dr. Ravi Subban · Zenodo (CERN European Organization for Nuclear Research) · 2018
Iris is one of important biometrics, which remains static and cannot be altered all-through the life of mankind. Iris is the most powerful biometric among all the other biological measures such as voice, fingerprint, palm print, signature and so on. Understanding the merit of iris, this paper presents an iris based security policy that grants access to the users by matching the iris of the individual with the trained iris sample. The main objective of this article is to attain maximum recognition accuracy. The objective is attained by segmenting the iris region and the extracted iris region is normalized. The Local Directional Pattern (LDP) features of the normalized iris images are extracted and the decision about user access grant or denial is made by the ensemble classifier. This work utilizes k-Nearest Neighbour (k-NN), Support Vector Machine (SVM) and Extreme Learning Machine (ELM) as ensemble classifier. The performance of the proposed approach is evaluated by varying the segmentation, feature extraction and classification techniques. On analysis, it is found that the performance of the proposed morphological operation based segmentation algorithm works better than the comparative segmentation algorithms. The idea of ensemble classification maximizes the recognition accuracy, as the final decision is made by considering the decisions of three efficient classifiers. The performance of the proposed approach is tested in terms of recognition accuracy, sensitivity and specificity. The proposed approach outperforms the existing approaches with better results.