Iris Recognition Using Multi-Algorithmic Approaches for Cognitive Internet of things (CIoT) Framework

Abstract

The recent widespread development of connected sensors, cloud, big data analytics, and ubiquitous sensing technologies have facilitated cognitive Internet of things (CIoT) and its emerging applications. Although CIoT has a great potential to affect human life, scholars have not explored how biometric technologies (e.g., iris) can contribute toward the success of CIoT-oriented framework, where iris-based biometric recognition is used for verification or authentication. One of the trade-offs of biometric recognition designs is to choose a unimodal- or multimodal-based structure. In this study, an iris-based recognition technology was developed as a unimodal biometric with the aid of multi-biometric scenarios. In the segmentation phase, a new algorithm based on masking technique to localize iris was proposed. Two new algorithms, namely, delta-mean and multi-algorithm-mean, were developed to extract iris feature vectors. The proposed system was evaluated on CASIA v. 1, CASIA v. 4-Interval, UBIRIS v. 1, and SDUMLA-HMT. Results show the satisfactory performance of the proposed solution for authentication issues.