Membrane-Inspired Bat Algorithm For Feature Selection To Recognize Faces In Unconstrained Scenarios

Loading...
Thumbnail Image
Date
2017-08
Authors
Alsalibi, Bisan A.N.
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Sains Malaysia
Abstract
Due to its non-intrusive nature, human analogy, and universal applicability, Face Recognition (FR) has gained significant attention in the last few decades. Notwithstanding the impressive developments in FR algorithms under controlled laboratory conditions, their performance may considerably degrade when moving to the wild. Intra-class variations such as illumination, occlusion, and expression can result in significant appearance changes. In this study, high-dimensional Local Binary Patterns were extracted from face images and fused with Gabor wavelet features using Canonical Correlation Analysis (CCA). To further enhance the discrimination power of facial representation, a novel membrane-inspired feature selection approach has been proposed, where a Binary Bat Algorithm (BBA) under the framework of Membrane Computing (MC) has been employed. Inherent parallelism and non-determinism are two distinguishing characteristics of MC that help in maintaining population diversity. In the proposed membrane-inspired BBA (MIBBA), the structure, evolution, dissolution and communication rules of MC were integrated into BBA to enhance bat trajectories. Furthermore, the Great Deluge Algorithm (GDA), has been integrated into the skin membrane to improve the exploitation ability of MIBBA. Experimental results using standard FR datasets show that the proposed approach yields competitive recognition rates and outperforms well-known state-of-the-art methods. Further experimental evaluations justified the ability of the proposed approach to handle the small sample size problem, where average recognition rates of 95.29%, 83.03% and 69.28% have been reported using AR, LFW and GBU databases, respectively.
Description
Keywords
Membrane-inspired bat algorithm for Feature selection , to recognize faces In unconstrained scenarios
Citation