wrong number of peaks using topology method.

[Tom89757]

Thank you first for your great library. I'm trying to find number of peaks. The code is below:

My origin image is like this:

fp.plot() shows:

fp.plot_mesh() shows:

It should have 6 peaks but only find 5 (with a wrong peak). Can you please explain it for me? I'll appreciate your help.

[erdogant]

I looked into it and it appears that the default noise reduction method (fastnl) is not optimal in your case.
If you take lee_enhanced, the peaks are as what we would expect by eye.

Update to the latest version with pip install -U findpeaks because I added a small function to read images from URL imread().

from findpeaks import findpeaks
path = r'https://user-images.githubusercontent.com/44827483/221152897-133839bb-7364-492a-921b-c9077ab9930b.png'
fp = findpeaks(method='topology', denoise='lee_enhanced', window=5, whitelist='peak')
X = fp.imread(path)

fp.fit(X)
fp.plot_persistence()
fp.plot()
fp.plot_mesh()

The persistence will give you more feeling how the peaks are detected. See more information here.

[Tom89757]

Thank you for your reply. It really helps me.

[erdogant]

👍🏻 closing this one as it is solved.