Prev: Filtering Noise from a signal
Next: designing a multiband filter
From: Tim Wescott on 2 Jun 2010 12:49
On 06/01/2010 05:40 PM, Tim Wescott wrote:
> On 06/01/2010 04:34 PM, bos1234 wrote:
>> In class we analysed a signal and filtered the noise out. However the
>> noise
>> was in a different bandwidth to the original signal hence it was easy to
>> filter.
>>
>> If noise and the signal were to overlap in freq. spectrum, are there any
>> techniques to filter out the noise??
>
> No. The word "filter" is used as a close analogy to a filter that you
> might use to filter liquids in the kitchen, or in a chemistry class.
> When you use a coffee filter filter, for example, you use a filter that
> has pores that are smaller than the coffee grounds. This lets the water
> and anything dissolved in it (like than nice coffee -- mmm!) through,
> but it blocks the coffee grounds because they won't fit.
>
> A filter in signal processing terms is much the same, except instead of
> filtering by size, you're filtering by position in the spectrum.
>
> Unless there is some characteristic of the signal that distinguishes it
> from the noise, you can't filter it out.
>
Actually, I shouldn't have given you a flat "no". Look into Wiener
filtering and Kalman filtering -- both of these are ways to find optimal
filters when the noise and signal spectra overlap. Both of them work
better the more you know about the signal and the noise (and both of
them can fail absurdly if your real signal and/or noise doesn't match
your assumptions -- but there are ways around that).

You're still playing a game where you're using what you know about the
signal and the noise to tease them out.

Note, also, that spread spectrum reception uses a time-domain "filter"
that will pull a signal out of -- apparently -- nothing; that's because
the underlying assumption of spectral analysis is that both signal and
noise are stationary, while spread spectrum techniques use a signal that
has considerable correlation over time, just in a way that doesn't show
up in a Fourier transform.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com
From: HardySpicer on 3 Jun 2010 03:51
On Jun 2, 11:34 am, "bos1234" <suren130(a)n_o_s_p_a_m.gmail.com> wrote:
> In class we analysed a signal and filtered the noise out. However the noise
> was in a different bandwidth to the original signal hence it was easy to
> filter.
>
> If noise and the signal were to overlap in freq. spectrum, are there any
> techniques to filter out the noise??

With some sort of extra information - perhaps! Normally we need either
a second signal + noise of some sort related to the first or some
measure of the noise on its own.

Hardy
From: illywhacker on 3 Jun 2010 06:12
On Jun 2, 6:49 pm, Tim Wescott <t...(a)seemywebsite.now> wrote:
> You're still playing a game where you're using what you know about the
> signal and the noise to tease them out.
>

This is the only game in town! This is more or less all of singal
processing.

illywhacker;
From: illywhacker on 3 Jun 2010 06:14
On Jun 2, 1:21 am, "bos1234" <suren130(a)n_o_s_p_a_m.gmail.com> wrote:
> In class we analysed a signal and filtered the noise out. However the noise
> was in a different bandwidth to the original signal hence it was easy to
> filter.
>
> If noise and the signal were to overlap in freq. spectrum, are there any
> techniques to filter out the noise??

Google 'denoising'. There is a vast literature on this. Of course, you
need to know *something* about your signal and your noise.

illywhacker;
From: Mikolaj on 3 Jun 2010 06:32
Dnia 02-06-2010 o 04:47:19 Fred Marshall <fmarshallx(a)remove_the_xacm.org>
napisa³(a):

(...)
> Well .... what about a line canceller? It uses a reference signal,
(...)

You assume a line and you cancel it.
You assume a noise and you cancel it.

If your assumption is proper
than substraction also will be proper.

I would look for noise shaping and Kalman filtering.
If the nois is stationary than there would be no need
to use adaptive assumptions for it.

--
Mikolaj
First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4
Prev: Filtering Noise from a signal
Next: designing a multiband filter