Finding index (x value) of noisy data crossing a threshold value

40 views (last 30 days)
Hello, I have some "noisy" data (xth_col) as shown by the red curve. I smooth this (black curve) first using:
win=200
d1=smooth(double(xth_col(:)),win);
I want to find out the width of this smoothed data by finding the x value where it crosses the horizontal red dotted line (=d80)
I have used:
xData=d1;
Value=d80
diff=abs(xData(:))-Value
[xshift, ind] = min(abs(diff(:)));
But this doesn't always work (as shown by the green data) . im going to need the first crossing coming from either end of the data. so I thought If I just list the diff, I can then find out where this swaps sign (i.e. crosses the horizontal line).
Not sure how to do this or if this is the best approach.
Thanks for any help
Jason

Accepted Answer

Jakob B. Nielsen
Jakob B. Nielsen on 4 Dec 2019
Edited: Jakob B. Nielsen on 4 Dec 2019
Have you considered the find function? It will give you every index of the input array that meets a certain condition.
cross=find(d1 > yourthreshhold);
cross(1) %yields the first index of xth_col which is above your threshhold
cross(end) %yields the last index of xth_col which is above your threshhold
  2 Comments
Jason
Jason on 4 Dec 2019
Thanks to both of you. I've accepted this answer only because it was first

Sign in to comment.

More Answers (1)

ME
ME on 4 Dec 2019
Edited: ME on 4 Dec 2019
idx = find(SmoothedData > Threshold);
Width = Xdata(max(idx)) - Xdata(min(idx));
Here the first line finds all indices of points above some threshold value. Then the second line calculates the time spent above this threshold by finding the difference between the X data points associated with the maximum and minimum indices.
Obviously, you'll need to adjust this to match your particular variable names, etc.
Is that what you wanted?

Categories

Find more on Particle & Nuclear Physics in Help Center and File Exchange

Tags

Community Treasure Hunt

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

Start Hunting!