Hi,
you can certainly use circular convolution to prevent
signal expansion after filtering. say your input is of
length L = 5 and the filter is of length M = 3. Then
generate a new signal by perodically extending the
input by length M-1 = 2:
input: 1 2 3 4 5
periodically extended input : 4 5 1 2 3 4 5 1 2
(length = 9)
now after doing convolution, you should ignore first
two values in the o/p. The next L values are your
filtered result, corresponding to the original input 1
2 3 4 5.
instead of a mere periodic extension, you may want to
use symmetric peridoc extension of the form:
3 2 1 2 3 4 5 4 3
For further details, check the following paper:
B. Usevitch, A tutorial on Modern Lossy Wavelet Image
Compression: Foundations of JPEG 2000, IEEE Signal
Processing Magazine, Sep 2001, pp 22-35.
hope that helps.
regards,
Sameer
---
"To know the road ahead, ask those coming back."
- Chinese proverb
--- ShadowsEdge Admin <shadowsedge@shad...>
wrote:
> Greetings,
>
> I was wondering if there was a way to perform
> convolution without lengthening the signal. Can
> cirular convolution be used for this? Everything
> I've looked up so far only mentions graphics
> operations. I am working with sound.
>
>
>
>
________________________________________________________________________
Yahoo! India Matrimony: Find your life partner online
Go to: http://yahoo.shaadi.com/india-matrimony
Reply by ShadowsEdge Admin●May 7, 20052005-05-07
Greetings,
I was wondering if there was a way to perform convolution without lengthening
the signal. Can cirular convolution be used for this? Everything I've
looked up so far only mentions graphics operations. I am working with sound.