Extracting (clearing) useful true data from a noisy temperature signal

Hi everyone, I have a temperature sensor signal with peak values. Sensor was setted for 0-275 K and 150-900 for two step of measurement but real temperatures were really high. I purposed to measure cooled points with my sensor between the peaks. Thus, temperature curve contains cooled points between the peaks. Many piece of frame after peaks, temperature getting down slowly relative to increasing of peaks and this cooled temperature trend is changing till the last frame.My point is acquiring this cooled temperature points between peaks. I called high temperature peaks as maximum pekas and low temperature peaks as min peaks. I tried findpeaks, localmax commands but I have to come up with a smart algorithm because peak widths are changing firstly at 381000th frame, secondly 445500th frame. I also attached data file with figure version of data file. How can I extract this cooled points along the frame? Are there anyone who can help me??

Thanks.

Here the link of .mat and .fig file;

https://easyupload.io/m/ajy5nq

I tried all combinations of findpeaks, locmin or locmax commands.

[Image Analyst]

You should attach the .mat file here. Make it easy for us to help you. We don't want to go to some third party web site.

[mehmet alper]

Of course right now. Just could not find the attachment button.

[mehmet alper]

.mat file can be easily plot with plot command by the way

Answer 1

[Image Analyst]

Looks like you want to extract data only where the local range in y values is low. Here's a start:

s = load('temps3-combinedloandhi.mat')

s = struct with fields:
    sumlohi: [1547847×1 double]

t = s.sumlohi;
x = 1 : numel(t);
subplot(3, 1, 1);
plot(x, t, 'b.-')
title('Original Data')
grid on;
% Filter data
windowWidth = 181;
tMax = movmax(t, windowWidth);
tMin = movmin(t, windowWidth);
tRange = tMax - tMin;
subplot(3, 1, 2);
plot(tRange, 'b-');
grid on;
title('Plot of the local y ranges')
% Plot histogram of local ranges.
subplot(3, 1, 3);
histogram(tRange, 128);
grid on;
title('Histogram of local range')
% Find where range is low.
mask = tRange < 10;
% Extract only data where the local range is low.
x2 = x(mask);
t2 = t(mask);
subplot(3, 1, 1);
hold on;
plot(x2, t2, 'r-')

[mehmet alper]

Thank you for this kindly effort! Apologise for my late. Thats a different perspective. This can work for the first section of the curve but indeed what I need is the points just before the minimal peaks. I attached 3 pictures for different frame intervals for my desired and undesired points.

On the other hand, the case is much more complex. I need to extract "desired points" and multiple these points with a constant. Then I need to extract undesired points and also multiple those points but with a different constant. Then I need to detect their locations as frame number to have calibrated cooled temperatures for whole frame points at the end of the day. I have just asked for desired points firstly, because it can be a good starting point fro my final purpose as I mentioned; " calibrated cooled temp points". If you can help that would be great Sir.

[mehmet alper]

Anyone to help?

[Image Analyst]

Did you try findpeaks?

[mehmet alper]

yes I tried but It found my points as peaks from 250 to 275 degree value, because of the sensor scale. At that part of the curve, my points were reaching into the maximum measurement pointof sensor (max. peak points). If you lead me into something, I can also try. Have to acquire these points nowadays. So thank you for your kindly attention, I hope we can solve this.

[mehmet alper]

Need a serious help anyone who can help is out there dear friends??

[mehmet alper]

Anyway, I found a way with using findpeaks reversely. Thanks you for your interest dear Image Analyst...