Find all the peaks higher than this threshold and save the corresponding range in a variable.

I have two plots on the same graph. The blue one is the MaxPerRangeBin and the red one is CfarThold.

I want to locate the peaks in blue which are higher than the red plot.

I have used islocalmax to locate all the peaks in MaxPerRangeBin , which is the logic I have to apply. Now, I have to just find out all the peaks which are just greater than red plot i.e. CfarThold.

I am sharing the snippet, along with the graph.

       TF = islocalmax(data(i).MaxPerRangeBin,2);
        aa(num_values,:) = data(i).MaxPerRangeBin(num_values,:);
        peakvalue_MPRB = aa(TF(num_values,:));
        

[Voss]

Can you please share the data and code you used to create the graph?

[Sara Nasir]

files = dir('*.mat');
data = struct();
for i = 1:length(files)
  result = load([files(i).folder,'\', files(i).name]);
  
    for num_values= 1:length(result.data)
        data(i).scanindex(num_values,:) =( result.data(num_values).ScanIndex);
        data(i).MaxPerRangeBin(num_values,:) = result.data(num_values).MaxPerRangeBin;
        data(i).CfarThold(num_values,:) = result.data(num_values).CfarThold;
        % extracting peaks
        TF = islocalmax(data(i).MaxPerRangeBin,2);
        aa(num_values,:) = data(i).MaxPerRangeBin(num_values,:);
        peakvalue_MPRB = aa(TF(num_values,:));
    end
    plot(data(i).MaxPerRangeBin(1,:));  % Blue plot
    hold on;
    plot(data(i).CfarThold(1,:));       % Red plot

end

[Image Analyst]

Again, can you attach the data? You forgot to attach any .mat files with the paperclip icon.

Answer 1

[Star Strider]

Possibly:

idxv = 1:numel(data(i).MaxPerRangeBin(1,:));                        % Index Vector
Lv = (data(i).MaxPerRangeBin(1,:)) >= (data(i).CfarThold(1,:));     % Logical Vector
TF = islocalmax(data(i).MaxPerRangeBin(1,Lv));                      % Peaks Logical Vector
TFidx = find(TF);                                                   % Numeric Indices
PeakIndices{i} = TFidx;                                             % Save Peak Index Values 

.

[Sara Nasir]

This is what I have done so far. I have the peaks in all_peaks vector as one column but I want it to be 2 dimensional matrix as TF or Thold.

num_values is 399 (means 399 rows).

We can skip saving the arrays in data structure for now :)

[row,col] = find(MaxPerRangeBin > CfarThold);
    
    Thold = nan(num_values,125);
  
    for ii=1:length(row)
           
         Thold(row(ii),col(ii))= MaxPerRangeBin(row(ii),col(ii));
     
    end

    TF = islocalmax(Thold,2); 
    all_peaks= Thold(TF); !!
    sorted_peaks= sort(all_peaks,'descend');  

[Star Strider]

Since there may not be the same number of peaks in each field, I would use the code I posted (unless there is a reason it does not work and would need to be changed). Saving the peaks to a cell array is preferable.

If it turns out that there are the same number of peaks in each field, use the cell2mat function on ‘PeakIndices’ to convert it into a matrix. Change the ‘PeakIndices’ assignment to:

PeakIndices{i,:} = TFidx; % Save Peak Index Values

to make that possible.

.

[Sara Nasir]

Thank you Star Strider for the help. I was able to save the values in a structure.

all_peaks =  uu( TF(num_values,:));
sorted_peaks = sort(all_peaks,'descend');  
       
data(num_values).result =  sorted_peaks; % saving to struct
      
c = ismember(Thold, sorted_peaks);      % extracting x-values
[ data(num_values).scan_value, data(num_values).bins] = find(c);
        
        

[Star Strider]

As always, my pleasure!