Re: Using butterworth filter

[StandardOctaveUser]

Also, if what we are seeing is not 
a time series of the motion but the 
frequency spectrum of it then you 
would need windowing to remove them,
(zigzags)

Basically divide data into m many windows
(equal length sub signals) multiply them 
by hanning(Nsub) where Nsub is your 
sub signal length. then average all those 
sub vectors. Then do the transition to 
frequency spectrum. (via fft). 

Increase or decrease m to get the desired
effect. As m increases Nsub decreases and 
though much smoother you start to lose 
detail in the frequency spectrum. 

Note: Can use hamming instead of hanning 
        again give it a go with a vector of all 
        ones and see which one you need.



--
View this message in context: 
http://octave.1599824.n4.nabble.com/Filtering-Signals-according-to-an-Amplitude-Range-tp4647610p4647936.html
Sent from the Octave - General mailing list archive at Nabble.com.