A robust scheme for iris segmentation in mobile environment
Narsi Reddy, Ajita Rattani, Reza R. Derakhshani · 2016
With the advent of mobile ocular biometrics and ocular human-computer interactions (HCI), recent research has been focused on iris localization in the visible spectrum. Existing studies suggest that performance of the ocular HCI and biometric systems significantly degrades due to inaccurate iris segmentation, especially when operating in uncontrolled mobile environment which causes variations such as motion blur, specular reflection and illumination variation. This paper proposes a iris segmentation algorithm for visible spectrum which is based on the combination of K-means clustering and Daugman's integro-differential algorithm. Experimental investigation on the publicly available VISOB dataset prove the efficacy of the proposed approach. Experimental results show that iris segmentation increases 4 folds compared to Daugman's and 3.5 folds compared to Masek's methods. The proposed method also executes 5 times faster than Daugman's and 8 times faster than Masek's methods.