A hybrid method for classification of physical action using discrete wavelet transform and artificial neural network. (21st March 2021)

• Record Type: Journal Article
• Authors: Jana, Gopal Chandra; Swetapadma, Aleena; Pattnaik, Prasant Kumar
• Is Part Of: International journal of bioinformatics research and applications. Volume 17:Number 1(2021)
• Page Start: 25
• Page End: 37
• Publication Date: 2021-03-21
• Subjects: EMG -- electromyography -- wavelet analysis -- DWT -- discrete wavelet transform -- ANN -- artificial neural network -- classification; Bioinformatics -- Periodicals; 570.285
• Languages: English
• ISSNs: 1744-5485

Abstract

This paper proposes a method for physical action classification based on wavelet analysis and artificial neural network (ANN) from electromyography (EMG) signals. The physical action includes the person's normal action as well as aggressive action. During various types of physical actions, the EMG signals are recorded. Discrete wavelet transforms (DWT) with DB-4 wavelet is used for feature extraction from recorded EMG signals. Extracted features are given as input to the ANN-based classifier to distinguish between normal actions and aggressive actions. The hybrid approach using combination of ANN and wavelet shows significance increase in level of accuracy in classifying the physical action. Hence proposed method can be used to discriminate the physical actions ultimately helps in identifying persons mental state.