Help understanding findpeaks function qualifier units?

What are the units on things in find peaks, like MinPeakHeight, MinPeakProminence, Width, Threshold etc. Is it based on my units in the data and I can size these based on what I see on the Y and x axes, or is it based on like pixels or something? I ask because prominence and threshold seems to be all over place. Amd you can't one side threshold for example at least X higher than data to the right, versus the left .

Answer 1

[Star Strider]

The dependent ‘y’ variable units are whatever they are in your data.

The independent ‘x’ variable units are either the indices (that I find most useful) or the original x vector units, if you supply them.

Admittedly, findpeaks does not do well with wandering baselines, so one way of correcting for that is to use a highpass filter to eliminate the d-c or constant offset and low-frequency baseline variations. (First do a fft (link) on your data to find the best cutoff frequency for your filter, then design the filter.)

[no nobody]

are there built in highpass filter functions I can use?

[Star Strider]

No. You have to design them yourself. They are relatively straightforward to design from the Signal Processing Toolbox functions, and easy with the designfilt (link) function.

Another option is the detrend (link) function, if all your trends are linear. (I prefer the filtering approach, because I have more control over it.)

[no nobody]

Okay I will look into it. How do I get findpeaks to spit out the results from this:

findpeaks(Peakdata,'MinPeakHeight',30,'MinPeakDistance',50,'MinPeakProminence',10,'Annotate','extents')

SO i can know what the prominence, width, location, and height of all of my peaks are?

[Star Strider]

You can only get what findpeaks lets you have. See the documentation in the section on Peak Prominences (link).

For your findpeaks call:

[pks,locs,widths,proms] = findpeaks(Peakdata,'MinPeakHeight',30,'MinPeakDistance',50,'MinPeakProminence',10,'Annotate','extents');

Without data that would work with your findpeaks call, I cannot run it to verify that those arguments will return anything useful in the outputs.

However with different data and a slightly different findpeaks call:

figure
findpeaks(Peakdata,'MinPeakHeight',10,'MinPeakDistance',10,'MinPeakProminence',1,'Annotate','extents');
axChild = get(gca, 'Children')

yields:

axChild = 

    4×1 graphics array:

    Line    (HalfProminenceWidth)
    Line    (Prominence)
    Line    (Peak)
    Line    (Signal)

you can then query those with, for example:

Half_Prom_Wdth = axChild(1)

and so for the others.

I’m not certain if this is more information than findpeaks returns in the outputs it lets you have. Nevertheless, the data appear to be there if you want to get them.