<abstract xmlns="http://www.w3.org/1999/xhtml"><p>Discrete wavelet transform (WT) followed by principal component analysis (PCA) has been a powerful approach for the analysis of biomedical signals. Wavelet coefficients at various scales and channels were usually transformed into a one-dimensional array, causing issues such as the curse of dimensionality dilemma and small sample size problem. In addition, lack of time-shift invariance of WT coefficients can be modeled as noise and degrades the classifier performance. In this study, we present a stationary wavelet-based two-directional two-dimensional principal component analysis (SW2D<sup>2</sup>PCA) method for the efficient and effective extraction of essential feature information from signals. Time-invariant multi-scale matrices are constructed in the first step. The two-directional two-dimensional principal component analysis then operates on the multi-scale matrices to reduce the dimension, rather than vectors in conventional PCA. Results are presented from an experiment to classify eight hand motions using 4-channel electromyographic (EMG) signals recorded in healthy subjects and amputees, which illustrates the efficiency and effectiveness of the proposed method for biomedical signal analysis.</p></abstract>

Discrete wavelet transform (WT) followed by principal component analysis (PCA) has been a powerful approach for the analysis of biomedical signals. Wavelet coefficients at various scales and channels were usually transformed into a one-dimensional array, causing issues such as the curse of dimensionality dilemma and small sample size problem. In addition, lack of time-shift invariance of WT coefficients can be modeled as noise and degrades the classifier performance. In this study, we present a stationary wavelet-based two-directional two-dimensional principal component analysis (SW2D2PCA) method for the efficient and effective extraction of essential feature information from signals. Time-invariant multi-scale matrices are constructed in the first step. The two-directional two-dimensional principal component analysis then operates on the multi-scale matrices to reduce the dimension, rather than vectors in conventional PCA. Results are presented from an experiment to classify eight hand motions using 4-channel electromyographic (EMG) signals recorded in healthy subjects and amputees, which illustrates the efficiency and effectiveness of the proposed method for biomedical signal analysis.

Stationary Wavelet-based Two-directional Two-dimensional Principal Component Analysis for EMG Signal Classification

School of Electrical & Information Engineering

School of Instrumentation Science & Opto-Electronics Engineering

ARC Centre of Excellence for Mathematical & Statistical Frontiers

Measurement Science Review

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Discrete wavelet transform (WT) followed by principal component analysis (PCA) has been a powerful approach for the analysis of biomedical signals. Wavelet...

Stationary Wavelet-based Two-directional Two-dimensional Principal Component Analysis for EMG Signal Classification

Online veröffentlicht: 15. Juni 2017

Seitenbereich: 117 - 124

Eingereicht: 17. Dez. 2016

Akzeptiert: 15. Mai 2017

DOI: https://doi.org/10.1515/msr-2017-0015

Schlüsselwörter
Wavelet transform, principal component analysis, feature extraction, pattern classification, electromyographic signal

© 2017 Yi Ji et al., published by De Gruyter Open

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Stationary Wavelet-based Two-directional Two-dimensional Principal Component Analysis for EMG Signal Classification

Online veröffentlicht: 15. Juni 2017

Seitenbereich: 117 - 124

Eingereicht: 17. Dez. 2016

Akzeptiert: 15. Mai 2017

DOI: https://doi.org/10.1515/msr-2017-0015

SchlüsselwörterWavelet transform, principal component analysis, feature extraction, pattern classification, electromyographic signal

© 2017 Yi Ji et al., published by De Gruyter Open

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Schlüsselwörter
Wavelet transform, principal component analysis, feature extraction, pattern classification, electromyographic signal