Biometric security system

Abstract

This thesis, "Biometric Security System: Unimodal Identification Using Finger Veins," explores the development and application of finger vein identification as a secure and efficient unimodal biometric recognition method. Leveraging advanced deep learning models, including InceptionResnet-V2 and a hybrid Convolutional Transformer-based approach (FVCT), the research establishes the potential for enhanced security and accuracy in biometric systems. Methodology involved the customization of deep learning architectures for finger vein identification, utilizing transfer learning and fusion of convolutional and transformer paradigms. Key findings demonstrate the superiority of these models, showcasing lower error rates and exceptional performance in comparison to existing state-of-the-art methods. Finger vein identification emerges as a reliable solution for diverse applications, from security to access control. Implications of these findings signify a path toward more secure and efficient biometric security systems. The fusion of deep learning paradigms and advancements in local feature extraction hold the promise of further innovation in the field. This research contributes to the ongoing development of robust and reliable personal identification solutions, ensuring enhanced security in critical domains.