Ant colony optimisation for latent fingerprint matching

Abstract

Biometric recognition is a prominent tool to recognise individuals based on their personal and biological traits like fingerprint, iris, voice, and face. Fingerprint recognition has been substantially used over the decades to identify and verify an individual's identity. In forensic investigations, fingerprint matching is one of the most reliable tools for person identification. The latent fingerprints collected from crime scenes are matched with full fingerprints for person identification. Latent fingerprints are accidentally left finger impressions with overlapping patterns, backgrounds and spoiled minutiae information. This paper aims to propose a system for automated latent fingerprint matching. The latent fingerprint images are initially enhanced to remove noise and to obtain useful information. Image enhancement includes pre-processing steps of segmentation followed by normalisation, filtering and image binarisation. Further, minutiae features are extracted from the pre-processed data and final matching is performed using an ant colony optimisation (ACO) algorithm to optimise the process of minutiae matching. The experimentation results on NIST special database 27 are computed in terms of accuracy assessment measures (precision, recall, F-score), similarity score and identification rate. The proposed system has evinced satisfactory results in comparison with other existing fingerprint matching techniques.