Classification of Hand Movements based on Discrete Wavelet Transform and Enhanced Feature Extraction

• Published in: International journal of advanced computer science & applications Vol. 10; no. 6
• Main Authors: Too, Jingwei, Rahim, Abdul, Mohd, Norhashimah
• Format: Journal Article
• Language: English
• Published: West Yorkshire Science and Information (SAI) Organization Limited 2019
• Subjects: Algorithms; Classification; Discrete Wavelet Transform; Discriminant analysis; Electromyography; Feature extraction; Machine learning; Pattern recognition; Rehabilitation; Support vector machines; Wavelet transforms
• ISSN: 2158-107X; 2156-5570
• DOI: 10.14569/IJACSA.2019.0100612

Extraction of potential electromyography (EMG) features has become one of the important roles in EMG pattern recognition. In this paper, two EMG features, namely, enhanced wavelength (EWL) and enhanced mean absolute value (EMAV) are proposed. The EWL and EMAV are the modified version of wavelength (WL) and mean absolute value (MAV), which aims to enhance the prediction accuracy for the classification of hand movements. Initially, the proposed features are extracted from the EMG signals via discrete wavelet transform (DWT). The extracted features are then fed into the machine learning algorithm for classification process. Four popular machine learning algorithms include k-nearest neighbor (KNN), linear discriminate analysis (LDA), Naïve Bayes (NB) and support vector machine (SVM) are used in evaluation. To examine the effectiveness of EWL and EMAV, several conventional EMG features are used in performance comparison. In addition, the efficacy of EWL and EMAV when combine with other features are also investigated. Based on the results obtained, the combination of EWL and EMAV with other features can improve the classification performance. Thus, EWL and EMAV can be considered as valuable tools for rehabilitation and clinical applications.