WA Multiscale Peak Detection VI

[ghighuphu]

Dear users,

using "WA Multiscale Peak Detection VI', I would expect that peaks detected for a waveform Y are the same as valleys detected for the waveform -Y (y-values multiplied by -1, i.e., waveform flipped upside-down), see the snippet, please. This is not the case, when doing the analysis on some data.

Opening and examining the subVIs, I only see reversing the synthesis and analysis filter frequencies and changing "greater" for "lower", as for the difference between "peaks" and "valleys" setting.

The documentation does not state any substantial difference between "peaks" and "valleys".

Any thoughts on it?

[mikeporter]

Just a thought. Is your data centered on zero?

Mike...

[ghighuphu]

Yes, y-offset was removed.

It seems that the data is unsymmetric or distorted... Needs more data :)

[ghighuphu · Accepted solution]

OK, so after some more analysis:

One has to "y-flip" the threshold for the valleys. I.e., if one searches for peaks with a threshold T and want to search for similar looking valleys, then there are two solutions:

a) flip the data (*(-1)) and keep the threshold

b) flip the threshold (*(-1)) and keep the data.

Having no option for searching the valleys offered by this VI, I'd came up with the a) solution quickly, most probably 🙂 -- it's hypothetical, right? Having this options and seeing the results, confused me highly.