Find peaks for a given 2D curve.

The code is very simple and needs only two lines of code.

Full call:

PeakValues=FindPeaksSimple(x,y,PeakType,Interpolate)

INPUT:

1 x: x-values. Vector with numbers.

2 y: y-values. Vector with numbers.

Optional values:

You may omit all off them or the trailing value.

3 PeakType: -1: only minima

0: minima and maxima (default)

1: only maxima

4 Interpolate: true or 1: parabolic interpolation

false or 0: no interpolation (default)

If the curve is noisy, you might consider to smooth it beforehand,

e.g. with https://mathworks.com/matlabcentral/fileexchange/66099

OUTPUT:

PeakValues: Array with peak x-positions and peak y-values

Empty PeakValues for no peak at all.

Processing time on my PC is:

No interpolation: 40us overhead + 17ns per sample.

Is for 10^6 samples: 17 ms.

With interpolation: 40us overhead + 17ns per sample + 34ns per peak.

Is for 10^6 samples with 3.4*10^5 peaks=28 ms

If you process always in the same manner, you can delete many code lines.

E.g. you want only nearest maxima:

   PeakValues=FindPeaksSimpleMaxima(x,y)
     PeakIndx = find(diff(sign(diff(y)))<0)+1;
     PeakValues=[x(PeakIndx)',y(PeakIndx)'];

Or you want only maxima interpolated:

  PeakValues=FindPeaksSimpleMaximaInterpolated(x,y)   
     PeakIndx = find(diff(sign(diff(y)))<0)+1;
     y1=y(PeakIndx-1);%Left of peak
     y2=y(PeakIndx);%At peak
     y3=y(PeakIndx+1);%Right of peak
     PeakPos=(x(2)-x(1))*(y3-y1)./(4*y2-2*y1-2*y3)+x(PeakIndx);
     PeakMinMax=y2-(y3-y1).^2./(8*(y1-2*y2+y3));
     PeakValues=[PeakPos(:),PeakMinMax(:)];