0
$\begingroup$

I want to know if there is a generalized way to select a signal from a set of signals. Let me clarify it: I have a number of signals $S = \{S1, S2, S3,...\} $.

Some of the signals are not important to me like, all having all zeros for data, or all same values, or only one datum comes in at the first occurrence then remains zero until the end, or the signal which is always a exactly/mostly straight line parallel to the time axis.

I made some heuristics but I am actually looking for a general scientific and technical methodology to do that. So, in fact I want to have a function $f(S) \rightarrow S1$, which takes a set $S$ and outputs $S1$ (a subset of $S$) after excluding S-S1 signals. Is there any thing in signal processing or does anyone know some paper/methodology to do this?

Improve this question
$\endgroup$
3
  • $\begingroup$ The definition of "important" or "meaningful" is quite application specific... I'm sure there are applications in which a flat line is a meaningful event worth being considered! There is nothing wrong with using heuristics like the one you describes as long as they embody knowledge about your application. $\endgroup$
    – pichenettes
    Dec 12, 2013 at 14:58
  • $\begingroup$ thanks! understandable. One more thing, what is the mthod in SP to know if a particular signal is how much flat or dense? what can you name it? because signal filtering I have checked does not solve this issue. I need to calculate a value from a signal which tells me how much(%) of the signal is flat or dense. thanks in advance! $\endgroup$
    – safdar aslam
    Dec 12, 2013 at 15:58
  • $\begingroup$ For this application something as simple as the energy of the derivative (approximated as the difference between consecutive samples) could work. Or more generally any measure of the energy in the highest frequency. Or maybe something like the spectral centroid $\endgroup$
    – pichenettes
    Dec 12, 2013 at 17:22

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.