how to filter data by rate of change?

I am trying to filter signal, where data of certain rate oif change is filtered out. I am not able to achieve my goals, for example with lowpass (tried for example lowpass(yTemp, 10, Fs2). below is plot with original signal to be filtered (blue) and with red, I have hand drawn what how I would like to filtered signal to behave. Of course it doesent need to be precisely like drawn, its just for example, to illustrate the point.

data is in attachment (x,y,Fs).

all help is appreciated

Answer 1

[Mathieu NOE]

hello

I was thinking that I could use the "baseline" correction approach, then apply a scale factor on the baseline compensated signal , add some smoothing then put back the base

and this is the result so far :

load('vars.mat')

% option # 1 : obtain the lower envelope data
N = 200;
[up1,down1] = envelope(yTemp,N,'peak'); 

% now apply smoothing on baseline corrected data and put back the baseline in
yc = yTemp - down1;
yscale_factor = 0.25; % apply only a fraction of yc (baseline corrected signal)
yTemps = smoothdata(yscale_factor*yc,'gaussian',2*N) + down1;
figure
plot(x,yTemp,x,down1,'--',x,yTemps,'linewidth',2)
legend('signal','lower envelope','smoothed');

[Mathieu NOE]

this is another alternative using the attached function :

% option # 2 : Baseline correction
[Base, yc]=baseline(yTemp(:));
figure
subplot(2,1,1),plot(x,Base,'--',x,yTemp,'linewidth',2)
subplot(2,1,2),plot(x,yc,'linewidth',2)
% now apply smoothing on baseline corrected data and put back the baseline in
N = 300;
yscale_factor = 0.25; % apply only a fraction of yc (baseline corrected signal)
yTemps = smoothdata(yscale_factor*yc,'gaussian',N) + Base;
figure
plot(x,yTemp,x,yTemps,'linewidth',2)
legend('signal','smoothed');

[Sven Larsen]

Thank you so much, this was precisewly what I was looking for. Both solutions are very good and it may be, I need to use lower envelope instead of smoothed.

Again, thanks! Will learn envelope and smoothdata, so I can use these also in future

[Mathieu NOE]

my pleasure ! glad I could help

[Sven Larsen]

what is peak_stripping in your line:

[Baseline, stripping]=peak_stripping(S,n);

[Mathieu NOE]

sorry, forgot to give you that function too

see attached :)