Uma nova métrica de qualidade de imagem para digitais infantis

Abstract

Biometrics is an identification technology that utilizes unique physical or behavioral characteristics of individuals. Its applications are wide-ranging, including security in electronic devices, access control in restricted areas, identity verification in financial transactions, and healthcare systems for medical record verification and vaccination control. Recently, the use of biometrics for children has been explored, especially in contexts such as schools and, primarily, hospitals. For the efficient use of biometric fingerprint systems, an image quality assessment step is necessary, where low-quality images can be discarded and recollected. Traditional fingerprint quality metrics, developed for adults, are not suitable for evaluating infant fingerprints, as they cannot differentiate between good and poor infant fingerprints. In a systematic literature review, no metric for use with infant fingerprints was found. Through the adaptation of NFIQ2, adding the ability to assess high-resolution images and considering the unique characteristics of infant fingerprints, this work achieved improvements in the evaluation of children’s fingerprint images. Experimental results demonstrated that the new metric outperformed NFIQ2, ensuring differentiation between high and low-quality infant fingerprint images, and maintaining the same evaluation capability in cases where NFIQ2 works.