So a 1D mean filter of size two with a sliding window defined as W below could be considered as having a stride of one.

Wn=(Xn + Xn+1) / 2


where as if I were to define a 1D 'stepped mean filter' with a stride of two and labeled U below.

Un= (X2n + X2n+1) / 2

where the number of samples in U (and also V) would be half that of W.

My questions are the following

  1. Is U still a filter (I'm guessing it is even though not fitting with the traditional kernel convolution implementation I associate with filters)
  2. What is the technical language one could use to communicate and research questions regarding filter/non-filter such as just given?
  3. All though not directly related but since I have the floor is, if V and U would produce equivalent results (my intuition says no)?
  • 1
    $\begingroup$ Google “multirate filters” “multirate signal processing” $\endgroup$
    – user28715
    Oct 16, 2019 at 14:56

1 Answer 1


That's a decimating filter. As such, it serves as a 1/2-rate resampler.


Your Answer

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

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