CARD-LESS ATM USING FINGERPRINT AND FACE RECOGNITION TECHNIQUES

In the field of biometric security, this study presents a Card-less ATM system leveraging fingerprint and face recognition techniques with a focus on Convolutional Neural Networks (CNN) and deep learning, implemented in the MATLAB domain. The methodology involves the acquisition of a dataset comprising fingerprint and face samples, followed by their training through a CNN model. During testing, both fingerprint and face images must belong to the same individual for authentication success, and a message box displays, granting access for transactions. In the case of a mismatch between the provided images or when they correspond to different individuals, the system responds with an "Authentication failed. Please try again." message. The system's accuracy relies on the efficient pre-processing of images, including resizing, ensuring reliable recognition and reducing false positives. This research contributes to enhancing ATM security by combining multiple biometric modalities to verify the user's identity, ultimately improving the accuracy and reliability of the authentication process.

Keywords: Fingerprint and face dataset, preprocessing, Splitting, Validation Security systems, Convolutional neural networks, Deep Learning, classification and Accuracy.