دانلود رایگان مقاله انگلیسی پوشش سیگنال الکترونیکی الکترومیوگرافی مبتنی بر تقسیم بندی دوتایی - الزویر 2018

abstract

In this work, we introduce two novel methodologies to compute the envelope of superficial electromyography signals. Our methods are based on the detection of activation and deactivation patterns using a change-point approach on the variances of the sample. More concretely, an iterative algorithms is proposed to select the change-points between two segments of the signal based on some local statistics introduced in this work. The signal is split up into two segments, and a new search for change-points is recursively conducted in each subsequence. The change-points make possible to calculate local envelopes which reflect the shape of the signal without ignoring the activation and deactivation landmarks. Two methods are proposed in this work, and the improvements with respect to methodologies available in the literature are shown using both synthetic and real data. A thorough analysis of the techniques is performed to that end.

5. Conclusions

In this work, we have proposed two novel methods (named MAbs and CIbs)to retrieve the shape and the amplitude ofthe bursts of sEMG signals. The proposed methods identify the start and end of bursts when other traditional methods cannot. Our comparisons were performed against the techniques of moving averages, rootmean squares and Butterworth low-pass filters using both real and synthetic signals. In particular, it is worth pointing out that the shape calculated through CIbs gets an estimation of the amplitude for each point in the signal.

It is important to point out that the methods proposed in this manuscript consider the presence of some quantitative parameters that may be tuned up to retrieve the shape, the peaks, the valleys and the jumps required in a wide range of EMG signal processing applications. In particular,the envelope can be as smooth as desired without losing the precision of the localization of the start and end of bursts. The methods were thoroughly validated using a wide range of experiments with real and synthetic data.