Grayscale Moving Average

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Grayscale Moving Average

Todd Johnson-4
I am new to ImageJ.  I have an image of numerous grayscale circles close
together on a black background.  I'd like to average the grayscale over
a 2x2 circle array (rectangular area plus background).  Then move to the
side one circle and repeat, obtaining a moving average of the average
grayscale values along the horizontal axis.  Then move down one circle
and obtain another horizontal axis data set, etc.  Finally, I'd like to
obtain grayscale values along the vertical axis in the same method.
What is the easiest way?
 
Thanks, Todd Johnson
 
Reply | Threaded
Open this post in threaded view
|

Re: Grayscale Moving Average

H. Gluender
>I am new to ImageJ.  I have an image of numerous grayscale circles close
>together on a black background.  I'd like to average the grayscale over
>a 2x2 circle array (rectangular area plus background).  Then move to the
>side one circle and repeat, obtaining a moving average of the average
>grayscale values along the horizontal axis.  Then move down one circle
>and obtain another horizontal axis data set, etc.  Finally, I'd like to
>obtain grayscale values along the vertical axis in the same method.
>What is the easiest way?
>
>Thanks, Todd Johnson
>

Dear Todd Johnson,

if I understand you correctly, you want to do what is called lowpass
filtering of an image by convolving it with the described kernel.

For doing so, just define the kernel in "Process > Filters >
Convolve..." and click ok. The resulting image then consists of the
running averages you are looking for.

HTH
--


                   Herbie

          ------------------------

          <http://www.gluender.de>
Reply | Threaded
Open this post in threaded view
|

Re: Grayscale Moving Average

Justin McGrath
In reply to this post by Todd Johnson-4
In the description he says he wants to move over one circle at a time.
 The convolution filter in ImageJ would move one pixel at a time
wouldn't it?  I think this wouldn't work well, since the only pixel
values one would want would be the ones where the kernel actually
lines up with the circles.

I'm not sure what values you are trying to get.  Do you want one
average for each horizontal "row", where each row consists of two rows
of circles? Or do you want a number for the first group of 2x2
circles, another number for the second, etc?

Either way, I think there is a particle analyzer that can find and
measure the mean gray scale value of each circle and put them into the
results.  Then it would be simple to put it into a spreadsheet and get
whatever numbers you're looking for.

You could also write a special convolution function where each step
was the size of the distance between circles, but this only works if
you know before hand precisely where the circles are in each image,
and I think the particle analyzer would be easier anyhow.

Justin

On 10/26/06, H. Gluender <[hidden email]> wrote:

> >I am new to ImageJ.  I have an image of numerous grayscale circles close
> >together on a black background.  I'd like to average the grayscale over
> >a 2x2 circle array (rectangular area plus background).  Then move to the
> >side one circle and repeat, obtaining a moving average of the average
> >grayscale values along the horizontal axis.  Then move down one circle
> >and obtain another horizontal axis data set, etc.  Finally, I'd like to
> >obtain grayscale values along the vertical axis in the same method.
> >What is the easiest way?
> >
> >Thanks, Todd Johnson
> >
>
> Dear Todd Johnson,
>
> if I understand you correctly, you want to do what is called lowpass
> filtering of an image by convolving it with the described kernel.
>
> For doing so, just define the kernel in "Process > Filters >
> Convolve..." and click ok. The resulting image then consists of the
> running averages you are looking for.
>
> HTH
> --
>
>
>                    Herbie
>
>           ------------------------
>
>           <http://www.gluender.de>
>