ORIGINAL ARTICLE
Cross-view gait recognition through ensemble learning
Xiuhui Wang
1
•
Wei Qi Yan
2
Received: 10 January 2019 / Accepted: 9 May 2019 / Published online: 16 May 2019
Ó Springer-Verlag London Ltd., part of Springer Nature 2019
Abstract
Gait has been well known as an unobtrusive promising biometric to identify a person from a distance. However, the
effectiveness of silhouette-based approaches in gait recognition is diluted due to variations of view angles. In this paper, we
put forward a novel and effective method of gait recognition: cross-view gait recognition based on ensemble learning. The
proposed method greatly enhances the effectiveness and reduces the sensitivity of gait recognition under various view
angles conditions. Furthermore, in this paper we will introduce a novel algorithm based on ensemble learning for com-
bining several gait learners together, which utilizes a well-designed gait feature based on area average distance. Through
experimental evaluations o n the well-known CASIA gait database and OU-ISIR gait database, our paper demonstrates the
advantages of the proposed method in comparison with others. The contribution of this research work is to resolve the
multiview angles problem of gait recognition through assembling several gait learners.
Keywords Gait recognition Cross-view Ensemble learning Gait classification
1 Introduction
Gait is composed of complicated human actions integrated
movements of body parts and joints together. As an
effective biometric [1, 2], human gait has a great deal of
advantages over classical biometrics, such as human face,
fingerprint, iris. (1) Unobtrusiveness. Most of the existing
biometrics require direct contact, specific postures or fixed
emotions for pattern classifications, while gait recognition
generally works in a natural way which does not need
special notifications for identifying persons. (2) Gait ima-
ges and videos are usually captured from a very far dis-
tance. Unlike face recognition, gait recognition utilizes
vision techniques to identify a person with a very low
resolution. Thus, gait recognition is a simple, attractive,
and effective method. (3) Gait images and videos are easily
taken using normal cameras like the one set forth in a
mobile phone. Even with low-quality images, gait recog-
nition algorithms can still work [3].
However, gait recognition remains a challenging prob-
lem in real applications, namely most of the exis ting gait
recognition algorithms only work well under the best
condition of image and video acquisition [4]. Corre-
spondingly, the performance of gait recognition algorithms
is dimmed due to varying illumination, view angles, human
clothing, etc. In this paper, a novel identification approach
of cross-view gait recognition based on ensemble learning
(CVGR-EL) is proposed. The CVGR-EL method takes use
of a novel gait representation, i.e., gait features based on
area average distance (AAD), which enhances the effec-
tiveness and reduces the sensitivity to variation s of various
view angles conditions. In this paper, we will develop a
novel algorithm to ensemble several learners and achieve
gait recognition.
The contributions of this research work are summarized
as follows: (1) A well-designed gait feature representation.
Based on area average distance, we propose a novel feature
representation and correspo nding feature extraction method
that makes the best use of intrinsic characteristics of human
gaits and improves gait recognition rate. (2) A novel view-
This work was supported in part by the National Natural
Science Foundation of China (NSFC) under Grant Nos.
61303146 and 61602431 as well as a scholarship from the
China Scholarship Council (CSC).
& Xiuhui Wang
wangxiuhui@cjlu.edu.cn
Wei Qi Yan
wyan@aut.ac.nz
1
China Jiliang University, No. 258, Xueyuan Street,
Hangzhou 310018, China
2
Auckland University of Technology, No. 2-14, Wakefiled
Street, Auckland 1010, New Zealand
123
Neural Computing and Applications (2020) 32:7275–7287
https://doi.org/10.1007/s00521-019-04256-z
(0123456789().,-volV)(0123456789().,-volV)