Overlapped Fingerprint Separation using Graph based Model

Abstract

For over a century, fingerprint analysis has been used to solve crimes, identify and track offenders. A common problem during the identification of a fingerprint is overlapped portion identification. This work aims to separate an overlapped fingerprint first and then identify the two separated fingerprints. In this approach, the parameters from a fingerprint are extracted and then represents with a graph based model using an adjacency matrix. Neighbour-finding algorithm has been applied to the matrix to track the ridge bifurcations. The proposed work aims to suggest the matrix representation of a sample with overlapping varies by a deviation factor from a non-overlapping segment. As a result, it can be made a concerted effort to sort out the overlapped fingerprints before relying on fingerprint matching in any significant way. The proposed graph-based approach successfully separated the overlapped portion and then match both the fingerprints.