I am still learning MATLAB and signal processing so not good at it. I have tried to be as much mature and upto the mark as possible but if there is something left so accept my apology for that.
*• I need to find range of only those Flat areas where I have local maximas like attached in the figure.
• I need to find the range of values where blue signal is flat near local maxims’. I have also make rectangular boxes around flat parts containing local maxima’s to get a clear view what I am expecting.*
Please have a look at the attached figure below.
I have tried to make a signal that replicates exactly my original data but unfortunately remained unable to come up with some good example so I have attached original data under link. I have applied one method on test example but unfortunately it’s not working with my original data.
Link for original Data Values:
https://www.mediafire.com/#7720j3w7b8pj4
Code to access plausible values of original data from link :
load originalflatenninroughnew t original_signal_data measured_Signal_data
[p l] = findpeaks(measured_Signal_data);
[pn ln] = findpeaks(-measured_Signal_data);
figure(1)
hold on
plot(t,original_signal_data,'g')
plot(t,measured_Signal_data,'b')
plot(t(l),p,'ko','MarkerFaceColor','r');
plot(t(ln),-pn,'ko','MarkerFaceColor','r');
legend ('originalsignal', 'measureddatasignal')
hold off
Detecting the flat regions in the blue signal
% create a random signal
rng ('default');
x = cumsum(rand(200, 1)-0.5);
% construct measured signal
x_measured = x;
x_measured (20:100) = 3.5; % first measured flat region
x_measured (105:135) = 2; % second measured flat region
The flat blue region are characterised by the fact that they have exactly the same value during the whole flat region. Therefore, one can detect the flat region by checking when two consecutive values are the same:
diff(measured_Signal_data)==0
The only problem is that the small flat regions are undesired, but we can remove them using an erosion and dilation step:
SE = strel('cube',20); % remove all flat regions smaller than 20 samples
flat = imdilate(imerode(diff(measured_Signal_data)==0, SE), SE)
The start and end indexes of the flat regions can be calculated by determining the rising/falling edges of flat:
% calculate the start and end of each flat region (rising and falling edge of flat)
startFlat = find(flat(1:end-1) == 0 & flat(2:end) == 1) + 1;
endFlat = find(flat(1:end-1) == 1 & flat(2:end) == 0) + 1;
The result can be visualised as follows:
figure
hold on
plot (x_measured);
plot (flat);
Dear Asif,
welcome
In order to reproduce more accurately the original signal you have to sample it in time domain such that your sampling frequency fs must be greater equal the maximum frequency contained in the original signal.In fact you need to oversample it for better location of the distinguished points such as the maxima and minima.
The other important point every sample must be presented with sufficient number of bits, that is with sufficient resolution to resolve the small amplitude variations of the signal.
After that you can locate the maxima and minima by differentiation and amplitude classification.
Best wishes
I have worked in R in order to find local extremes and I have created R Package 'RootsExtremaInflections'.
See examples here:
https://www.researchgate.net/publication/317169647_RootsExtremaInflections_Demo
You need function extremexi() of that package.
If you could share data by another way, I'd probably try (I don't want to create an account for downloading data).
Presentation RootsExtremaInflections Demo
- Badges
- Science topic