Aplicação da técnica OMNI utilizando matching como medida de distância para redução do espaço de busca em bancos de dados de biometria digital infantil

Abstract

Fingerprint biometrics have proven effective in identifying newborns, helping to prevent baby swaps in hospitals and facilitating vaccine tracking. However, as databases of infant fingerprints grow, issues of search slowness during identification arise. Typically, the search is performed exhaustively, comparing each element in the database one by one. In very large databases, this can become a significant hindrance. Therefore, it is necessary to study methods to speed up the retrieval of these fingerprints, which is the focus of the present work. There are several studies on large-scale search improvements; however, to date, none have been tested on infant fingerprint databases. Thus, the objective of this work is to apply the OMNI technique, which shows a sub-linear reduction in search space relative to the database size. Fingerprint matching was used as a distance measure, specifically the MCC algorithm, something not yet tested for infant fingerprint biometrics. The experiments conducted showed reductions of up to 92% in search space, maintaining a 100% accuracy rate. Additionally, it was experimentally possible to use MCC as a distance measure in a metric space. For future work, it is recommended to test different matching algorithms and add more fingerprints to the database to verify the robustness and efficiency of the approach in these cases.