Thread Subject: Filter for an application

I have a question about filtering.
I am working on an experiment with mice brain slices and I
want to get the Spatial FITC fluorescence intensity
profiles.I have a time series of images taken from a CCD
camera. From those images I drew a line (in ImageJ
software) to get the intensity of some pixels, the plotted
data (pixels)seems to have Gaussian noise.
What is the best filter to get rid of that noise? I
implemented a moving average filter and the result looks
good but I want to make sure that the moving average filter
is the best option, if not could you guys please let me
know about it?

Thanks in advance for your attention

hg

"Hugo Caicedo" <hkaizer21@hotmail.com> wrote in message
<fnb63q$8l8$1@fred.mathworks.com>...
> I have a question about filtering.
> I am working on an experiment with mice brain slices and
I
> want to get the Spatial FITC fluorescence intensity
> profiles.I have a time series of images taken from a CCD
> camera. From those images I drew a line (in ImageJ
> software) to get the intensity of some pixels, the
plotted
> data (pixels)seems to have Gaussian noise.
> What is the best filter to get rid of that noise? I
> implemented a moving average filter and the result looks
> good but I want to make sure that the moving average
filter
> is the best option, if not could you guys please let me
> know about it?
>
> Thanks in advance for your attention
>
> hg

Is your noise distributed in space - i.e. present in a
single frame. Or in time - that is variation of a given
pixel between frames.

The answer to your question depends heavily on which type
you are trying to minimize.

Regards

Dave Robinson

Dave,
Thanks for answering me.
The noise is distributed in both space and time, actually I
have an image of it. Do you know how I could upload
additional information (like images .jpg) in this Newsgroup?

Thank you

hg

"Dave Robinson" <dave.robinson@somewhere.biz> wrote in
message <fncalq$2ge$1@fred.mathworks.com>...
> "Hugo Caicedo" <hkaizer21@hotmail.com> wrote in message
> <fnb63q$8l8$1@fred.mathworks.com>...
> > I have a question about filtering.
> > I am working on an experiment with mice brain slices
and
> I
> > want to get the Spatial FITC fluorescence intensity
> > profiles.I have a time series of images taken from a
CCD
> > camera. From those images I drew a line (in ImageJ
> > software) to get the intensity of some pixels, the
> plotted
> > data (pixels)seems to have Gaussian noise.
> > What is the best filter to get rid of that noise? I
> > implemented a moving average filter and the result
looks
> > good but I want to make sure that the moving average
> filter
> > is the best option, if not could you guys please let me
> > know about it?
> >
> > Thanks in advance for your attention
> >
> > hg
>
> Is your noise distributed in space - i.e. present in a
> single frame. Or in time - that is variation of a given
> pixel between frames.
>
> The answer to your question depends heavily on which type
> you are trying to minimize.
>
> Regards
>
> Dave Robinson

"Hugo Caicedo" <hkaizer21@hotmail.com> wrote in message
<fndn6q$fnl$1@fred.mathworks.com>...
> Dave,
> Thanks for answering me.
> The noise is distributed in both space and time, actually I
> have an image of it. Do you know how I could upload
> additional information (like images .jpg) in this Newsgroup?
>
> Thank you
>
> hg
>
> "Dave Robinson" <dave.robinson@somewhere.biz> wrote in
> message <fncalq$2ge$1@fred.mathworks.com>...
> > "Hugo Caicedo" <hkaizer21@hotmail.com> wrote in message
> > <fnb63q$8l8$1@fred.mathworks.com>...
> > > I have a question about filtering.
> > > I am working on an experiment with mice brain slices
> and
> > I
> > > want to get the Spatial FITC fluorescence intensity
> > > profiles.I have a time series of images taken from a
> CCD
> > > camera. From those images I drew a line (in ImageJ
> > > software) to get the intensity of some pixels, the
> > plotted
> > > data (pixels)seems to have Gaussian noise.
> > > What is the best filter to get rid of that noise? I
> > > implemented a moving average filter and the result
> looks
> > > good but I want to make sure that the moving average
> > filter
> > > is the best option, if not could you guys please let me
> > > know about it?
> > >
> > > Thanks in advance for your attention
> > >
> > > hg
> >
> > Is your noise distributed in space - i.e. present in a
> > single frame. Or in time - that is variation of a given
> > pixel between frames.
> >
> > The answer to your question depends heavily on which type
> > you are trying to minimize.
> >
> > Regards
> >
> > Dave Robinson
>

To make your life easy you need access to the Image
Processing Toolbox - if you have this then check out the
wiener2 ("help wiener2") this provides one very good way of
reducing the spatial noise in the image. An alternative, if
you have the Wavelet toolbox is detailed in the help there
try typing in "wavemenu" (my quote marks, don't type in
those) at the command prompt, and take it from there.

Your time distributed noise is slightly more difficult. A
running average does work - BUT what you must be wary about
is that the noise in your image is offset -i.e. you never
get a negative pixel, so with heavy averaging, the noise
doesn't integrate to zero, but to some positive number. This
has the effect of wrecking the contrast of the processed image.

I will give some thought to methods I have seen used, of
which there are quite a few. Again I would investigate a
wavelet based methodology. If you consider each pixel, it
will generate a single time distributed waveform, and each
of these waveforms are amenable to wavelet noise reduction
techniques.

Perhaps one of the Forum Guru's will have better ideas.

Regards

Dave Robinson