Hi there, I have a vector which is not evenly distributed in the time domain. But I still want to filter out the spikes on the curve using a LPF constructed from fft. What I want to do is to find the frequency where the largest amplitude in the frequency domain oppears, add 1 to it and use the outcome as the filter threshold.
But the thing is, I don't know how to calculate f without any info of time step. And apparently, if you give it different time steps you will get different base frequency, and the filtered signal can be quite different.
And my questions are:
- Can I use fft in this case?
- If so, how to implement it?
- If not, what are some workarounds for this?
I attached data files for better understanding. col.mat is the signal which is not evenly distributed in time. And the filtered signal should be something like the shape in BW.pdf.
Any help is really appreciated. And thanks in advance!