0
$\begingroup$

I would like to extract the location of data jump (see figure below), the difficulty is that the data are noisy as you can see. I try with a median filter or a other regularization and it could move the location of my jump.

Perhaps i could compute differences between consecutive data and then extract the most important one. Nonetheless my data could contain two jumps... And for information I m working with Python

enter image description here

$\endgroup$
  • 1
    $\begingroup$ Take a look at this related question and its answers. $\endgroup$ – Matt L. Jul 18 '18 at 10:54
  • $\begingroup$ Is something wrong with my answer? $\endgroup$ – Marcus Müller Sep 7 '18 at 11:46
2
$\begingroup$

Use a low-pass filter with linear phase. Since group delay is the derivative of phase over frequency, linear phase filters have constant group delay.

That way, yes, you shift that jump, but simply by subtracting the group delay, you get the original position.

Group delay for a discrete causal filter is half its length.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.