Gait Recognition System Using Gabor Wavelet and Active Gait Differential Image

Abstract

Gait is a behavioural biometric process that serves to identify persons using their walking style. It is un-obstructive, not easy to conceal and offers distance recognition. Various approaches have been employed to improve the performance and accuracy of gait biometric systems but the performance is yet to measure up to that of other biometric recognition systems. In this work, Gabor wavelets were used to extract Active Gait Differential Image (AGDI) features, while Principal Component Analysis (PCA) was used for feature dimensionality reduction. The classification was performed using Support Vector Machine (SVM) and silhouette images from Chinese Academy of Science Institute of Automation (CASIA) gait dataset was used for testing. The performance was evaluated using accuracy, equal error rate, false acceptance rate and false rejection rate and it gave 99.19%, 1%, 0%, and 2% respectively for the metrics used.