Bio-medical and latent fingerprint enhancement and matching using advanced scalable soft computing models

Latent fingerprints are acquired from crime places which are utilized to distinguish suspects in crime inspection. In general, latent fingerprints contain mysterious ridge and valley structure with nonlinear distortion and complex background noise. These lead to fundamentally difficult problem for further analysis. Hence, the image quality is required for matching those latent fingerprints. In this work, we develop a model for enhancement of latent fingerprint and matching algorithm, which requires manually marked (ground-truth) ROI latent fingerprints. This proposed model includes two phases (i) Latent fingerprints contrast enhancement using type-2 intuitionistic fuzzy set (ii) Extract the minutiae and Scale Invariant Feature Transformation (SIFT) features from the latent fingerprint image. For matching, these algorithms have been figured based on minutiae and SIFT points which inspect n number of images and the scores are calculated by Euclidean distance. We tested our algorithm for matching, using some public domain fingerprint databases such as Fingerprint Verification Competition − 2004 (FVC-2004) and Indraprastha Institute of Information Technology (IIIT)-latent fingerprint which indicates that by fusing the proposed enhancement algorithm, the matching precision has fundamentally moved forward.