MATLAB Answers

How do I cut the signal into several parts?

11 views (last 30 days)
Sh Sh
Sh Sh on 21 Oct 2021
Edited: Image Analyst on 22 Oct 2021
Welcome
I have 16-channel sEMG signals. I want to take the signal from each channel, which consists of a set of movements, including A rest movement is followed by a movement, then a rest movement, followed by the same movement, a second attempt, and so on until you complete 6 repetitions of one movement between each repetition and the last rest. I know the number of samples for each movement, so how can I remove the rests between iterations of the EMG signal, which represents zeros samples, and only get the repeats, each iteration alone

Accepted Answer

Jon
Jon on 21 Oct 2021
Edited: Jon on 21 Oct 2021
You can use logical indexing. So if you wanted to remove the portion of a signal x where it equaled zero you could do something like this
idl = x==0
xnew = x(idl)
or more compactly
xnew = x(x==0)
This just removes the zero portions. If you want individual vectors for each non-zero piece it would take a little more work
Here is a way to do it. It seems like there is probably a more elegant solution but I think this approach will work. Note it assumes that the only time the signal is exactly zero is when at rest. It also assumes that the data starts with movement.
% define an example vector with rest portions where value is zero
x = [1 2 3 4 0 0 0 5 6 9 0 0 2 3 8 0 0 4 9 5];
% get logical indices for non-rest portions
isMotion = x~=0;
% find transitions from rest to movement .. .these will have a value of 1 and -1
% respectively
jmp = diff([0 isMotion 0]); % pad with zeros to ensure transition at begin and end
% find beginning and end of movement segments
iStart = find(jmp == 1);
iEnd = find(jmp == -1)-1;
% assign segments to structure
% (can't put them in an array unless we know all motions have same length)
numSeg = numel(iStart);
xseg(numSeg).x = 0; % preallocate
for k = 1:numel(iStart)
xseg(k).x = x(iStart(k):iEnd(k));
end

More Answers (1)

Image Analyst
Image Analyst on 22 Oct 2021
Edited: Image Analyst on 22 Oct 2021
Please post the graph along with annotations about where you want to do things to the signal.
Did you try thresholding?
quietIndexes = y < someValue; % for example someValue could be 0 or 0.001 or whatever.
Then you can use bwlabel() to label quiet stretches or high signal stretches, take 6 of them using ismember(), and get their lengths with regionprops() (if you want).
labeledSignal = bwlabel(~quietIndexes);
first6 = ismember(labeledSignal, 1:6);
props = regionprops(first6, 'Area');
allLengths = [props.Area];
% Remove quiet parts so all signals run together
output = y(first6);

Tags

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!