Authors:
Chirawat Wattanapanich
and
Hong Wei
Affiliation:
University of Reading, United Kingdom
Keyword(s):
Gait, Gaussian, Entropy, SVM, PCA.
Related
Ontology
Subjects/Areas/Topics:
Applications
;
Biomedical Engineering
;
Biomedical Signal Processing
;
Biometrics
;
Biometrics and Pattern Recognition
;
Learning of Action Patterns
;
Multimedia
;
Multimedia Signal Processing
;
Pattern Recognition
;
Software Engineering
;
Telecommunications
Abstract:
This paper investigates the effect of lower knee gait representations on gait recognition. After reviewing three emerging gait representations, i.e. Gait Energy Image (GEI), Gait Entropy Image (GEnI), and Gait Gaussian Image (GGI), a new gait representation, Gait Gaussian Entropy Image (GGEnI), is proposed to combine advantages of entropy and Gaussian in improving the robustness to noises and appearance changes. Experimental results have shown that lower knee gait representations can successfully detect camera view angles in CASIA Gait Dataset B, and they are better than full body representations in gait recognition under the condition of wearing coat. The gait representations involving the Gaussian technique have shown robustness to noises, whilst the representations involving entropy provide a better robustness to appearance changes.