Browse by author
Lookup NU author(s): Dr Jie ZhangORCiD
This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).
Wearable sensor-based human activity recognition has been widely used in many fields. Consideringthat a multi-sensor based recognition system is not suitable for practical applications andlong-term activity monitoring, this paper proposes a single wearable accelerometer-based humanactivity recognition approach. In order to improve the reliability of the recognition system andremove redundant features that have no effect on recognition accuracy, wavelet energy spectrumfeatures and a novel feature selection method are introduced. For each activity sample, waveletenergy spectrum features of the acceleration signal are extracted and the activity is representedby a feature set including wavelet energy spectrum features and features of other attributes. Then,considering the limitation of single filter feature selection method, this paper proposes an ensemblebasedfilter feature selection (EFFS) approach to optimize the feature set. Features that are robustto sensor placement and highly distinguishable for different activities are selected. In the experiment,the acceleration data around waist is collected and two classifiers: k-nearest neighbour (KNN)and support vector machine (SVM) are utilized to verify the effectiveness of the proposed featuresand EFFS method. Experiment results show that the wavelet energy spectrum features can increasethe discrimination between different activities and significantly and improve the activity recognitionaccuracy. Compared with other four popular feature selection methods, the proposed EFFS approachprovides higher accuracy with fewer features.
Author(s): Tian Y, Zhang J, Wang J, Geng Y, Wang X
Publication type: Article
Publication status: Published
Journal: Systems Science & Control Engineering
Print publication date: 18/02/2020
Online publication date: 18/02/2020
Acceptance date: 19/01/2020
Date deposited: 18/02/2020
ISSN (print): 2164-2583
ISSN (electronic): 2164-2583
Publisher: Taylor & Francis
Altmetrics provided by Altmetric