Features extraction of fingerprints based on hybrid particle swarm optimization and bat algorithms

Abstract

Most security system's essential errand is to check that the people are in fact who they claim to be. In Contrast to traditional techniques such as passwords and smart cards that are used in some organizations, fingerprint identification may be preferred as it makes the information virtually impossible to steal. The most extensive used biometric features are Fingerprints, in order to identify a person because of their uniqueness and invariance. The fingerprint consists of valleys and ridges on the surface of a fingertip. In this paper, a new hybrid strategy Particle Swarm Optimization (PSO) with Bat Algorithm (BA) is proposed to extract features from fingerprint images. Both PSO and BA algorithms are swarm-based algorithms that mimics the swarm behaviour of particles and bats in nature. In the field of image processing, features are extremely significant. Before obtaining features, the noisy area should be removed from the foreground first, and then several important techniques are applied on each sample image in the database such as Fingerprint Enhancement by using Fast Fourier Transform (FFT), Binarization, and Thinning. The hybrid (PSO-BA) algorithm is proposed as a pre-enhancing step to select the clear minutiae (or feature) structures across several iterations, which will be more suited for the matching phase. By comparing the proposed method with several methods in calculating FAR and FRR, the results showed that the FAR (0.001) and FRR (0.01) were less than the other proposed methods. That means the hybrid (PSO-BA) algorithm has the better results, which means it can be used as one of the best search approaches to extract features from fingerprints