how to use property "ProminenceWindow" for command islocalmin

Can someone please explain how to use property "ProminenceWindow" for command islocalmin. I want to understand what difference this ProminenceWindow property can make.

My assumption is i can filter out peak/valley according to the width of the peak/valley(do not confuse width with 'halfprom' or 'halfheight' in WidthReference property of findpeak. I want to find full prominence width i.e X indices of start of peak and end of peak. using this width can be easily found by subtacting)

i can filter out peak/valley according to the height if i use MinPeakProminence property

Please check below example that i tried for understanding this property.

Example code:

% Generate sample data

x = linspace(0, 3*pi, 1000); % Extended range for gradual changes

y = sin(0.5*pi*x) + 0.01*randn(size(x)); % Half sine wave with very minimal noise

% Define different ProminenceWindow values

prominenceWindows = [5, 20, 50]; % Adjust these values to see the effect

% Create a figure with subplots

figure;

for k = 1:length(prominenceWindows)

% Detect local minima with the current ProminenceWindow

minIdx = islocalmin(y, 'ProminenceWindow', prominenceWindows(k));

% Count the number of local minima detected

numMinima = sum(minIdx);

% Create a subplot for the current ProminenceWindow

subplot(length(prominenceWindows), 1, k);

plot(x, y, '-b', 'DisplayName', 'Data');

hold on;

plot(x(minIdx), y(minIdx), 'ro', 'MarkerFaceColor', 'r', 'DisplayName', 'Local Minima');

hold off;

legend show;

title(['ProminenceWindow = ', num2str(prominenceWindows(k)), ', Count = ', num2str(numMinima)]);

xlabel('x');

ylabel('y');

end

% Adjust subplot layout for better visualization

sgtitle('Local Minima Detection with Different ProminenceWindow Values');

Answer 1

[Suraj Kumar]

Hi Swati,

The ‘ProminenceWindow’ is a property in the ‘islocalmin’ function which is used to define the range over which the prominence of each local minimum is calculated. Prominence measures how much a minimum stands out compared to its surroundings.

To visualize the effect of ‘ProminenceWindow’ parameter, we need to set the ‘MinProminence’ parameter as well as it specifies a threshold for the prominence of local minima. Only minima with a prominence equal to or greater than this value are considered.

Now without this, the count of minima remains constant regardless of the 'ProminenceWindow' setting since all local minima are considered, irrespective of their prominence values.

You can refer to the attached code snippet and the output for a better understanding:

x = 1:100; 
A = peaks(100); 
A = A(50,:); 
figure; 
% First subplot 
subplot(2, 1, 1);   
TF1 = islocalmin(A, 'MinProminence', 1, 'ProminenceWindow', [5, 5]); 
plot(x, A, x(TF1), A(TF1), 'r*'); 
title('Plot 1'); 
axis tight; 
% Second subplot 
subplot(2, 1, 2);   
TF2 = islocalmin(A, 'MinProminence', 2, 'ProminenceWindow', [15, 15]); 
plot(x, A, x(TF2), A(TF2), 'r*'); 
title('Plot 2'); 
axis tight; 

Output:

To know more about the ‘ProminenceWindow’ parameter in the ‘islocalmin’ function in MATLAB, you can refer to the following documentation:

• https://www.mathworks.com/help/matlab/ref/islocalmin.html#:~:text=ProminenceWindow%20%E2%80%94%20Prominence%20window

Hope this works for you!