dilation in 3D

[Jonathan Leong]

hello imagej users,

i am trying to perform a dilation on an image stack. the dilate and maximum
features built into imagej are able to do what i want, but only in 2D. that
is, they take the maximum value in an n-pixel radius - but within a given
image plane. is there a way to do this with an n-pixel radius in the entire
volume?

ideally, i would want to be able to specify the distance between slices
(i.e. the conversion from z steps to pixels), though i would be happy to
start with a macro/plugin in which i could simply dilate 3-pixels and that
would give me a 3-pixel radius spanning 3 slices. in fact, this is 1-pixel
=1 z-step is approximately correct.

i have done a bit of searching around and found the VTK plugins, but i am
having trouble getting these to work. they are quite old, though, so my hope
is that there is something newer that is native to imagej. does anyone have
any ideas?

thanks,

jonathan

[Thomas Boudier]

Hi Jonathan,

Maybe you can use my 3D filters :
http://imagejdocu.tudor.lu/doku.php?id=plugin:filter:3d_filters_with_jni:start

this is a jni version so it is quite fast, it uses a ellipsoid kernel
where you can specify xy radius and z radius.

Hope this helps,

Thomas

> hello imagej users,
>
> i am trying to perform a dilation on an image stack. the dilate and maximum
> features built into imagej are able to do what i want, but only in 2D. that
> is, they take the maximum value in an n-pixel radius - but within a given
> image plane. is there a way to do this with an n-pixel radius in the entire
> volume?
>
> ideally, i would want to be able to specify the distance between slices
> (i.e. the conversion from z steps to pixels), though i would be happy to
> start with a macro/plugin in which i could simply dilate 3-pixels and that
> would give me a 3-pixel radius spanning 3 slices. in fact, this is 1-pixel
> =1 z-step is approximately correct.
>
> i have done a bit of searching around and found the VTK plugins, but i am
> having trouble getting these to work. they are quite old, though, so my hope
> is that there is something newer that is native to imagej. does anyone have
> any ideas?
>
> thanks,
>
> jonathan
>
>
>  

--
/*****************************************************/
 Thomas Boudier, MCU Université Pierre et Marie Curie
 IFR 83 Biologie Intégrative. Bat B 709, Campus Jussieu
 Tél : [phone]  Fax : [phone]
/*****************************************************/