Volume 18, No. 3, 2021
Multimodal Biometric Fusion Of Face Iris In Person Recognition Framework
Asha K H , Manjunathswamy B E , Krishnamurthy M , Sunil Kumar G and Mustafa Basthikodi
Abstract
Advancement of biometric frameworks in real-time applications contains mostly unimodal biometric frameworks, where the information gathered from single trait. There is chance that single traits do not recognize a person rightly because of existence of some limitations with modalities we choose. By making use of multiple biometric modalities using fusion operation, those limitations are overridden. In this paper, a novel multimodal biometric person recognition system is proposed, that is based on two fusion methodologies feature-level and score-level, which are compared comprehensively, for face and iris traits, in order to predict the methodology which gives higher recognition rates, to determine system’s performance. We have identified four databases for face images and two databases for iris images and combined these in to eight groups of data sets for the experimentation, with four features extraction approaches such as GLCM, LBP, FD and PCA applied separately. The higher recognition rates are achieved for score-level fusion when used with the GLCM and LBP approaches, with the minimal EER and maximum accuracy. The performance of the proposed model is compared with existing multimodal face-iris biometric frameworks.
Pages: 213-230
Keywords: Multimodal Biometric, Face-Iris, Fusion, Feature-Level, Score-Level