Closed kuwisdelu closed 1 month ago
We need robust noise estimation in for N-D signals.
We have x(t) = f(t) + n(t) and want to determine n(t).
x(t) = f(t) + n(t)
n(t)
We can use a filter to estimate f^(t) from x(t), then estimate n^(t) = x(t) - f^(t).
f^(t)
x(t)
n^(t) = x(t) - f^(t)
Then we can summarize the noise n^(t) using sd(), mad(), quantile(), etc.
n^(t)
sd()
mad()
quantile()
What filter to use for estimating f^(t)? Gaussian? Mean? Median?
Initial draft of functionality added in https://github.com/kuwisdelu/matter/commit/a522beff61763b398ec43955e1da645bd14de470
We need robust noise estimation in for N-D signals.
We have
x(t) = f(t) + n(t)
and want to determinen(t)
.We can use a filter to estimate
f^(t)
fromx(t)
, then estimaten^(t) = x(t) - f^(t)
.Then we can summarize the noise
n^(t)
usingsd()
,mad()
,quantile()
, etc.What filter to use for estimating
f^(t)
? Gaussian? Mean? Median?