Iris-Based Biometric Authentication for Next-Generation ATM Security

With the rapid expansion of electronic banking services, securing Automated Teller Machine (ATM) transactions has become a critical concern for financial institutions. Conventional card-and-PIN authentication mechanisms are increasingly vulnerable to threats such as card skimming, shoulder surfing, PIN compromise, and identity theft. This paper proposes a next-generation ATM security framework based on iris biometric authentication integrated with a deep learning–driven classification model. The system captures a user’s iris image, preprocesses and encodes distinctive texture features, and performs template matching against a securely stored biometric database. The proposed architecture eliminates the need for physical cards and memorized PINs, thereby reducing attack surfaces associated with traditional systems. Experimental validation conducted using the IIT-Madras Iris Database demonstrates improved authentication accuracy and robustness compared to conventional authentication methods. The results indicate that iris-based verification offers enhanced security, usability, and resistance to spoofing, positioning the proposed solution as a viable model for next-generation ATM infrastructures. The rapid expansion of electronic banking services has significantly increased the demand for secure and reliable transaction systems. Conventional Automated Teller Machine (ATM) security mechanisms based on debit/credit cards and Personal Identification Numbers (PINs) are vulnerable to theft, card skimming, phishing, and shoulder-surfing attacks. To address these limitations, this paper proposes a next-generation ATM security system based on iris biometric authentication integrated with deep learning techniques. The proposed system captures the user’s iris image using a high-resolution infrared camera and verifies identity through template matching using a Convolutional Neural Network (CNN)-based classifier. The uniqueness and stability of iris patterns make them highly resistant to forgery and impersonation attacks. The system was evaluated using the IIT-Madras iris database, achieving improved accuracy and robustness compared to traditional authentication methods. The proposed solution eliminates the need for physical cards and PINs, offering enhanced security, usability, and reliability for future banking infrastructure.