Gaussian shape for 'findpeaks' function?

Hello everyone, probably this is an easy question…. for the ‘findpeaks’ function how can I do its plot looks like a Gaussian shape and modify the markers style? Thanks. Here is the code:

ax17=subplot(1,2,2);
[pks4,locs4,widths4,proms4] = findpeaks(pws4(:,end),y,'MinPeakHeight',25);  
findpeaks(pws4(:,end),y,'Annotate','extents','WidthReference','halfprom'); % Gaussian shape? how ??
text(locs4+0.5,pks4,num2str((1:numel(pks4))'));
legend('Filtered Data','Peak','Prominence','Width','FontSize',5);
ax17.YAxisLocation = 'right';
ax17.XDir = 'reverse';
ax17.XGrid = 'on';
ax17.YGrid = 'off';
title ('Time (20:30 - 21:00)');
ylabel('Layering identification','FontSize',9,'FontName','Arial','color','default');
xlabel('Altitude/km','FontSize',9,'color','default');
yticklabels('');
xlim(ax17,[75 95]);
ylim(ax17,[15 19]);
camroll(-90)

%typo edited

[Mathieu NOE]

hello

can you provide the parts mat file ?

[Fercho_Sala]

@Mathieu NOE , yes,, the file is attached.

[Mathieu NOE]

error because of this line

load parts;

Error using load

Unable to read file 'parts'. No such file or directory.

[Fercho_Sala]

@Mathieu NOE without that line it works (typo edited), the matter is how you can make a Gaussian shape for the whole peaks line.

[Mathieu NOE]

that's what I tried also , but I got this strange plot , left half empty and right half as below :

[Fercho_Sala]

@Mathieu NOE yes, the reason is that there is not a plot for the left side, this is another different code... and of course the limits must be resized, the question is more about to represent the line (data) in a Gaussian shape, it is possible?, Sir @Stephen Cobeldick any comment?

[Mathieu NOE]

hello

there are some points to be clarified :

• pws4 has 7 columns , but seems you are only interested in the last column - correct ? : pws4(:,end)
• this line does not work as the data has no peak above 25 (max value is about 22)

[pks4,locs4,widths4,proms4] = findpeaks(pws4(:,end),y,'MinPeakHeight',25);  

• the next line works but I don't understand what you are trying to achieve - what does the gaussian shape stuff mean ? you would like to fit a king of gaussian pulse to the data ? and take the gaussain peak value coordinates ?
• basically this is what this line gives :

but then I wondered if the question was about a technique to shape the data to give them a king of gaussian shape

if that is the question see the output of these two smoothing filters; the sgolay performs better IMHO

code :

% smoothing filters (makes the data have a pseudo gaussian shape)
figure(2)
tmp = pws4(:,end);
tmps = medfilt1(tmp, 25,'truncate');
tmpss = sgolayfilt(tmp,2,71);
plot(y,tmp,y,tmps,y,tmpss);legend('Raw','Smoothed medfilt1','Smoothed sgolayfilt');
title('Data Smoothed');

Answer 1

[Nipun]

Hi Fercho,

I understand that you want to plot the peaks found using the "findpeaks" function in a way that resembles a Gaussian shape and modify the marker style. Here's how you can do it:

1. Fit a Gaussian to the peaks.
2. Modify the markers in the plot.

Here is the modified version of your code:

% Example data
y = linspace(0, 100, 1000);
pws4 = randn(1000, 1) + 50*exp(-(y-50).^2/(2*10^2)); % Example data with peaks

% Find peaks
[pks4, locs4, widths4, proms4] = findpeaks(pws4, y, 'MinPeakHeight', 25);  

% Fit Gaussian to each peak
gaussEqn = 'a*exp(-((x-b)/c)^2)'; % Gaussian function

figure;
ax17 = subplot(1,2,2);
hold on;

% Plot original data
plot(y, pws4, 'DisplayName', 'Filtered Data');

% Plot peaks with Gaussian fit
for i = 1:length(pks4)
    % Fit Gaussian
    x_peak = y(locs4(i)-10:locs4(i)+10);
    y_peak = pks4(i) * exp(-((x_peak-locs4(i))./widths4(i)).^2);
    plot(x_peak, y_peak, '--r', 'LineWidth', 1.5); % Gaussian fit
    
    % Plot peak markers
    plot(locs4(i), pks4(i), 'xk', 'MarkerSize', 10, 'LineWidth', 2); % Marker style
end

% Annotate peaks
text(locs4 + 0.5, pks4, num2str((1:numel(pks4))'));

% Set plot properties
legend('Filtered Data', 'Gaussian Fit', 'Peak');
ax17.YAxisLocation = 'right';
ax17.XDir = 'reverse';
ax17.XGrid = 'on';
ax17.YGrid = 'off';
title('Time (20:30 - 21:00)');
ylabel('Layering identification', 'FontSize', 9, 'FontName', 'Arial', 'color', 'default');
xlabel('Altitude/km', 'FontSize', 9, 'color', 'default');
yticklabels('');
xlim(ax17, [75 95]);
ylim(ax17, [15 19]);
camroll(-90);

In this code:

1. We fit a Gaussian function to each peak and plot it.
2. We modify the marker style for the peaks.

Hope this helps.

Regards,

Nipun