Human gait recognition using firefly template segmentation

Abstract

ABSTRACT Gait recognition is one of the advanced behavioural biometric technology, which aims to distinguish people by their walking style. The unique advantage of gait recognition is its capability to catch gait at a distance without the prior consent of the subject. In the gait recognition process, the features of human motion are automatically extracted and later used to authenticate the identity of the person in motion. The template-based model-free gait recognition method offers an optimal solution for gait recognition through gait energy image (GEI) and gradient gait energy image (GGEI), which increases the performance of gait recognition against covariates like normal walking conditions, bag, and coat. In this paper, we propose a novel firefly template segmentation (FTS) method, which employs the firefly algorithm to accomplish the boundary selection process. The principal component analysis (PCA) technique is used for dimensional reduction and the multiple discriminant analysis (MDA) technique applies to achieve better class separability. The proposed work is tested on a publicly available CASIA-B dataset and the experimental results show excellent performance in comparison with other gait recognition methods reported in the literature.