How to extract informative part from a signal ?

Hi, I'm working on a dataset of EMG signals that comes from different muscles: biceps, quadriceps and so on... (globally 6 different classes). This signals are stored as vector of dim 1, n. I was wondering if there is the possibility to extract the central part of these signals automatically, I have already tried to compute the power in slot of 100 values and if a slot is greater than the threshold I keep it (but it's not working always, and not generalizable). I put an image in order to better understand which is my problem. Thank you

Answer 1

[yanqi liu]

yes，sir，may be use data filter to smooth data vector，and check the diff to segment data，get the split index range to segment

my be upload your data mat file to do some analysis

[Alberto Azzari]

Thank you for your answer, I have uploaded as an attachment a part of the dataset. Which kind of filter I need to impement ?

[yanqi liu]

clc; clear all; close all;
load test
x = ds(1,:);
x2 = x;
for i = 1 : 10
    x2 = smooth(x2,100);
end
[pks1,locs1,w1,p1] = findpeaks(x2,'MinPeakProminence',5e-3);
[pks2,locs2,w2,p2] = findpeaks(-x2,'MinPeakProminence',5e-3);
locs = [locs1; locs2];
figure; plot(x);
hold on;
plot([min(locs) min(locs)], [min(x) max(x)], 'r--');
plot([max(locs) max(locs)], [min(x) max(x)], 'r--');

[Alberto Azzari]

Thank you, it works