The problem of high Gabor features has led to some computational complexities in
Gabor-based facial recognition system. Several dimensionality reduction techniques such linear
subspace and non-linear approaches have been proposed to reduce high volume of Gabor features.
This paper applies a nature-inspired meta-heuristics optimization algorithm using Ant Colony
Optimization Algorithm to obtain relevant and optimal features from huge Gabor features. The
evaluation of the proposed system was achieved using two image datasets; Olivetti Research
Laboratory (ORL) database and African Face Image Database (AFI). The paper further applied
three distance classifiers; Malahanobis, Euclidean and Chebyshev to classify face image into
either matched or mismatched. Observations from the experimental study showed that Mahalanobis
has the highest classification accuracy of 97.14% with better False Acceptance Rate (FAR) obtained
in Mahalanobis and Chebyshev, better False Rejection Rate (FRR) in Chebyshev for AFI Database.
Also, the best classification accuracy of 95.71% was achieved in Mahalanobis with better False
Acceptance Rate (FAR) and False Rejection Rate (FRR) recorded both in Mahalanobis and Chebys
PAPER TITLE :ENHANCED GABOR FEATURES BASED FACIAL RECOGNITION USING ANT COLONY OPTIMIZATION ALGORITHM ARO, T.O.*; ABIKOYE, O.C.; OLADIPO, I.D. AND AWOTUNDE, B.J.
JOURNAL Of SUSTAINABLE TECHNOLOGY | VOLUME 10 NUMBER 1 2019
Paper Details
- Author(s) : ARO, T.O.*; ABIKOYE, O.C.; OLADIPO, I.D. and AWOTUNDE, B.J.
- Abstract:
