Design and implementation of an artificial neural network applied to finger bad-positioning detection on touchless multiview fingerprints devices
fingerprint, multiview, touchless, quality assessment
This paper presents a method based on Artificial Neural Network that evaluates the rotational bad-positioning of fingers on touchless multiview fingerprinting devices. The objective is to determine whether the finger is rotated or not, since a proper positioning of the finger is mandatory for high fingerprint matching rates. A test set of 9000 acquired images has being used to train, validate and test the proposed multilayer Artificial Neural Network classifier. To our knowledge, there is no definitive method that addressed the problem of fingerprint quality on touchless multiview scanners. The proposed finger rotation detection here presented is one of the steps that must be taken into account if a future automatic image quality assessment method is to be considered. Average results show that: (a) our classifier correctly identifies bad-positioning in approximately 94% of cases; and (b) if bad-positioning is detected, the rotation angle is correctly estimated in 90% evaluations.
Zaghetto, Cauê; Aguiar, Luiz Henrique Morais; Zaghetto, Alexandre; and Vidal, Flávio de Barros, "Design and implementation of an artificial neural network applied to finger bad-positioning detection on touchless multiview fingerprints devices" (2015). Proceedings of the XI Brazilian Symposium on Information Systems (SBSI 2015). 72.
This paper is in Portuguese (Projeto e implementação de uma rede neural artificial para detecção do mal-posicionamento rotacional de dedos em dispositivos de captura de impressões digitais multivista sem toque)