Iris Liveness Detection using Fusion of Thepade SBTC and Triangle Thresholding Features with Machine Learning Algorithms

Abstract

Conventional security systems are often plagued by inherent flaws, leading to frequent security breaches. To address these vulnerabilities, automated biometric systems have emerged, leveraging individuals' physiological and behavioural traits for precise identification. Among these biometric modalities, iris-based authentication is a highly reliable, distinctive, and contactless method for user recognition. This research endeavours to enhance the accuracy of iris liveness detection by combining features extracted from the TSBTC n-Ary (Thepade’s Sorted Block Truncation Coding) method with those derived from the Triangle Thresholding method. Two distinct datasets, namely IIIT Delhi and Clarkson 2015, have been employed to evaluate the efficacy of these combined features. The study involves extracting features from three sources: TSBTC, TSBTC+Triangle, and Triangle methods. These features are subsequently input into the WEKA tool, which employs various classifiers to assess accuracy. The findings of this investigation reveal a notable increase in the accuracy of Iris Liveness Detection (ILD) by incorporating handcrafted techniques like TSBTC in conjunction with the Thresholding method. In essence, this research underscores the potential for improving the robustness of security systems by harnessing the synergy of distinct biometric methods, thereby mitigating the shortcomings of conventional security systems and fortifying the foundations of secure user authentication.