What code to implement to create a non-causal moving average filter that would use future values as input?
조회 수: 7 (최근 30일)
이전 댓글 표시
I am quite confused as I am able to write a code for a simple causal moving average filter but do not know that what would differ if it's a non causal one. Please provide me the code for non-causal moving average filter for any number of M values.
댓글 수: 0
답변 (2개)
Image Analyst
2015년 10월 24일
See https://en.wikipedia.org/wiki/Causal_filter "In signal processing, a causal filter is a linear and time-invariant causal system. The word causal indicates that the filter output depends only on past and present inputs. A filter whose output also depends on future inputs is non-causal, whereas a filter whose output depends only on future inputs is anti-causal. "
If your signal is time, with early times on the left and later times on the right, then I guess a non-causal filter would be a window that had an odd number of elements and didn't have any on the past/left side of it, like
kernel = [0,0,0,0,1,1,1,1,1] / 5;
out = imfilter(signal, 1, kernel);
or something like that.
댓글 수: 3
Image Analyst
2015년 10월 25일
No, it only averages over half the filter window. To do a moving average over the whole window, which is what you're asking now, you'd use all 1's.
y = conv(x, ones(1, M)/M, 'full');
hawi aboma
2021년 3월 27일
편집: hawi aboma
2021년 3월 27일
I'm confused how to write the matlab code for and design filter for cousal and non cousal system, and how to check real time filtert?
댓글 수: 1
Image Analyst
2021년 3월 27일
What's confusing? For a sliding window filter, half of it should be zeroed out - the half that refers to the "future".
참고 항목
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!