Peak finder - finding major peaks instead of just the smaller ones

I want to use a peak finder to find the major peaks in my spectrum, however, when I use {pks, locs] = peakfinder()

I get all the small peaks as well. I just want the main five

Answer 1

[Greg Dionne]

You can do:

findpeaks(y,'NPeaks',5,'SortStr','descend')

This gets you the five with the largest amplitude. If you want the five with the largest prominence, you can do:

[pks,locs,~,prm] = findpeaks(y);
[~,i] = sort(prm,'descend');
plot(1:numel(y),y,'-',locs(i(1:5)),y(locs(i(1:5))),'o ')

[Russ]

how would you do this to find the five minimum peaks?

i.e. within acceleration data, the most prominent negative peaks

Answer 2

[Azzi Abdelmalek]

minimum_peak=10 % For example
[pks,locs]=findpeaks(your_signal,'MINPEAKHEIGHT',minimum_peak)

[Bran]

yea I thought of using minpeakheight but my peaks aren't very strong as you can probably see from the plot I included as an attachment. For some reason my peaks are not too good. i think it because I have a large spike at 0Hz which is an error. I dont really know how to get rid of it though. It may be due to g as my data is from an accelerometer?

[Azzi Abdelmalek]

You need some criterion to get rid of unwanted peaks

[José-Luis]

You could also transform your data prior to detecting the peaks. Maybe a log transform would be enough?