Prof. Arti Tekade, Mustafa Baldiwala,Aditya M Deshmukh ,Prof. Maithili Andhare, Kountey Wadje, Prof. Vijayalakshmi Kumbhar

Biometric radiographs have gained increasing significance in recent times due to the surge in crime and disaster incidents. Presently, authenticating and identifying individuals has become a fundamental component of most computer vision automation systems. Traditional biometric methods such as fingerprints, iris scans, facial recognition, and palm prints fall short when it comes to recognizing individuals whose external biometric features have been compromised due to conditions like rashes, injuries, or severe burns. Ensuring security, robustness, privacy, and resistance to forgery are pivotal aspects of any person authentication system. In such challenging scenarios, utilizing radiographs of the skull, hand, and teeth emerges as effective alternatives. The below research introduces an innovative approach for human authentication based on forensic hand radiographs, employing a deep neural network. The feature extraction from hand radiographs and subsequent recognition tasks are executed through a three_x0002_layered convolutional deep neural network architecture. In our experimentation, we analyzed a dataset comprising hand radiographs obtained from subjects of varying age groups, professions, and genders. The algorithm’s performance is rigorously assessed through cross-validation accuracy, wherein we systematically vary parameters such as striding pixels, pooling window size, kernel size, and the number of filters. Our experiments unveil the presence of biometric information within hand radiographs, enabling the identification of individuals, particularly in the context of disaster victim identification.

https://doi.org/10.58357/jedcs.2024.v3.i2.19