An Approach for the Preprocessing of EMG Signals Using Canonical Correlation Analysis
EMG signals are generally contaminated by various kinds of noises in a heterogeneous way. Among these various noises, major issue is the proper removal of Additive White Gaussian Noise (AWGN), whose spectral components overlay the spectrum of EMG signals;
- PDF / 223,100 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 85 Downloads / 260 Views
Abstract EMG signals are generally contaminated by various kinds of noises in a heterogeneous way. Among these various noises, major issue is the proper removal of Additive White Gaussian Noise (AWGN), whose spectral components overlay the spectrum of EMG signals; making its analysis troublesome. This paper presents an approach for AWGN removal from the EMG signal using Canonical Correlation Analysis (CCA). In this approach, CCA is first performed on the noisy EMG signals to break them into various canonical components followed by Morphological Filtering. Herein, a square-shaped structuring element is deployed which filters the canonical components. After that, the outcomes of the proposed methodology are contemplated with the approaches adopted in CCA-Gaussian filtering and CCA-thresholding. Outcomes of simulations show that the preprocessing approach used in this work suppresses AWGN from EMG signal while preserving the original content.
1 Introduction Electromyogram (EMG) is an electrical revelation originated as a result of muscle contraction. The procurement of a pure EMG signal forms a major issue for proper analysis and utilization of the signal [1]. At low contraction level, the surface EMG D. Anand (&) V. Bhateja A. Srivastava D.K. Tiwari Department of Electronics and Communication Engineering, Shri Ramswaroop Memorial Group of Professional Colleges (SRMGPC), Lucknow 226028, Uttar Pradesh, India e-mail: [email protected] V. Bhateja e-mail: [email protected] A. Srivastava e-mail: [email protected] D.K. Tiwari e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2018 S.C. Satapathy et al. (eds.), Smart Computing and Informatics, Smart Innovation, Systems and Technologies 78, https://doi.org/10.1007/978-981-10-5547-8_21
201
202
D. Anand et al.
signal is usually contaminated by three categories of noises: PLI, Baseline Wander, and AWGN [2]. AWGN originates as a consequence of many recording electrodes and their electrode skin contact area, which is nowadays a more challenging issue. Several preprocessing approaches have been developed in past for the removal of AWGN from EMG signals [3]. Among these, the most common technique is Gaussian filtering for suppression of AWGN. However, Gaussian filter also eliminates useful biomedical information from this signal. Hence, effective removal of AWGN without deteriorating the signal quality is a challenging task in preprocessing of EMG signals. Amirmazlaghani and Amindavar [4] introduced an approach of EMG noise suppression based on statistical modeling of wavelet coefficients using GARCH Modeling; but GARCH model often fails to capture highly irregular phenomena. Aschero and Gizdulich [5] carried out noise removal using Modified Wiener Filtering. However, this approach can only handle processes with additive, unimodal noise. Sobahi [6] propounded an approach of wavelet-based filtering of the EMG signals which targeted every AWGN component according to a thresholding rule. In continuation to previous work, Veer and Aga
Data Loading...
