Gait recognition based on wavelet features with spectral regression kernel discriminant analysis

Abstract

This paper has proposed gait recognition approach for analyzing and classifying human identification under carrying a bag and wearing a clothing thus improving recognition performances. The proposed method is based on detail wavelet features extracted from the Haar-wavelet decomposition of dynamic areas in the Gait Energy Image (GEI). Spectral Regression Kernel Discriminant Analysis (SRKDA) is then applied to the extracted feature vector in order to reduce its dimensionality by selecting only the most relevant and discriminate features. The CASIA Gait database under variations of clothing and carrying conditions for different viewing angles was used to evaluated the proposed method using a k-NN classifier and has yielded an attractive performance of up to 93% in terms of identification Rate (IR) at rank-1 when compared to similar and existing state-of-the-art methods.