Survey on the convergence of learning techniques for precise pattern identification in fingerprint analysis

Abstract

Fingerprint analysis has been an essential component of forensic science and biometric authentication systems. The precise recognition of fingerprint patterns is critical for solving crimes, confirming identities and assuring the security of numerous applications. In recent years, the convergence of learning approaches, particularly machine learning and deep learning, has played an essential part in improving the accuracy and efficiency of fingerprint pattern recognition. In this paper, we discuss recent advancements in physiological-based biometric multimodalities, with a particular focus on the field of precise pattern identification in fingerprint analysis. Additionally, we assume the task of summarizing and examining a range of physiological-based biometric modalities, encompassing both traditional and deep learning approaches. A detailed review of several biometric measurements across many modalities is presented, encompassing various phases, including preprocessing, feature extraction and classification, which are thoroughly discussed. We provide a comprehensive analysis of the challenges and future developments associated with conventional and deep learning methodologies. The objective is to enable investigators to recognize these problems. A review is conducted to evaluate the standard and deep learning approaches employed in different physiological-based biometric systems. The comparative analysis of this review suggests that more advancement is necessary for the development of a reliable physiological-based approach to enhance and optimize the functionality of the fingerprint system.