Forums

Low complexity approach of measuring signal/image local activity?

Started by walala January 24, 2004
Dear all,

I am facing with the following image/signal processing problem:

In order to measure image local activity, I take variance. But
variance involves square operation, suppose I am working on NxN
blocks, then the N^2 square will need a lot of computation, which is
very costly in my current application. Is there any techniques of
measuring local activity by way of lower complexity computations?

Thanks a lot,

-Walala
"walala" <mizhael@yahoo.com> wrote in message
news:6f348bd1.0401241748.18ec006e@posting.google.com...
> Dear all, > > I am facing with the following image/signal processing problem: > > In order to measure image local activity, I take variance. But > variance involves square operation, suppose I am working on NxN > blocks, then the N^2 square will need a lot of computation, which is > very costly in my current application. Is there any techniques of > measuring local activity by way of lower complexity computations?
How about taking the sum of the magnitude differences from the mean or the difference between the max and the min? Those measures might take more computation than you want also.... FFT and sum the high frequency components. Fred
"Fred Marshall" <fmarshallx@remove_the_x.acm.org> wrote in message
news:UOqdnQwNHLbd7I7dRVn-gQ@centurytel.net...
> > "walala" <mizhael@yahoo.com> wrote in message > news:6f348bd1.0401241748.18ec006e@posting.google.com... > > Dear all, > > > > I am facing with the following image/signal processing problem: > > > > In order to measure image local activity, I take variance. But > > variance involves square operation, suppose I am working on NxN > > blocks, then the N^2 square will need a lot of computation, which is > > very costly in my current application. Is there any techniques of > > measuring local activity by way of lower complexity computations? > > How about taking the sum of the magnitude differences from the mean or the > difference between the max and the min? Those measures might take more > computation than you want also.... > > FFT and sum the high frequency components.
... I might better better said "magnitudes" here. Without that the sums can be zero or very low.