What is the meaning of "width" and "prominences" in findpeaks

It said that:

[pks,locs,w,p] = findpeaks(data) additionally returns the widths of the peaks as the vector w and the prominences of the peaks as the vector p.

w is the widths of the peaks as the vector w

I still unable to find out how they calculate the width.

Example link

In the example below:

x = linspace(0,1,1000);

Pos = [1 2 3 5 7 8]/10;
Hgt = [4 4 2 2 2 3];
Wdt = [3 8 4 3 4 6]/100;

for n = 1:length(Pos)
    Gauss(n,:) =  Hgt(n)*exp(-((x - Pos(n))/Wdt(n)).^2);
end

PeakSig = sum(Gauss);

Plot the individual curves and their sum

plot(x,Gauss,'--',x,PeakSig)
grid

findpeaks(PeakSig,x,'MinPeakProminence',2,'Annotate','extents','WidthReference','halfheight')
[pks,locs,w,p]= findpeaks(PeakSig,x,'MinPeakProminence', 2,'Annotate','extents','WidthReference','halfheight')
title('Signal Peak Widths')

w =

0.2352 0.1725

p =

4.8721 3.0028

Is the w is the width (half-height) and p is height as pointed in the plot?

Answer 1

[Star Strider]

Is the w is the width (half-height) and p is height as pointed in the plot?

Your interpretation of ‘p’ is correct. The value of ‘w’ returned depends on how you define 'WidthReference' (link). You can define it to be full width half maximum (FWHM) if you define it as 'halfheight'.

[Matlaber]

Thanks! I didn't see the documentation in details.

I had tried:

findpeaks(PeakSig,x,'MinPeakProminence',2,'Annotate','extents','WidthReference','halfprom')
[pks,locs,w,p]= findpeaks(PeakSig,x,'MinPeakProminence', 2,'Annotate','extents','WidthReference','halfprom')
title('Signal Peak Widths')

w =

0.2351 0.1724

p =

4.8721 3.0028

(The W is almost the same, different by 0.0001)

W = width (half-prominence for 'halfprom' default)

W = width (half-height for 'halfheight')

[Star Strider]

My pleasure.

It is quite possible that ‘prominence’ and ‘height’ are essentially the same in some situations.

[lakshman s]

Is it possible to get the endpoints (data points) of the widths apart from the distance?

[Duy Tran]

is it possible to get the endpoints (data points) of the widths apart from the distance?

[Star Strider]

It would be necessary to know the half-prominence height (‘y’ value that is apparently half the ‘y’ value of the tallest peak) and then use interp1 to calculate at least one of the ‘x’ values, then adding or subtracting the FWHM value to get the second ‘x’ value. (That could be something of a challenge in the examples in this illustration.)