Help to use the Parallel Spectral Deconvolution (Piotr Wndykier)

[Dani Díaz de Quijano-2]

Dear all,

I am a PhD student at the University of Barcelona, at the Department of
Ecology. I am new in ImageJ, and I am interested in restoring 3D widefield
images of bacteria and algae for intensity quantification purposes.

I found the plugin Parallel Spectral Deconvolution (Piotr Wndykier) very
interesting.

I would need real restoration algorithms (constrained iterative or
statistical image restoration, p.e, adn NOT no neighbour, closest neighbour
or linear algorithms). Are the implemented algorithms (Generalized Tikhonov,
Tikhonov and TSVD image deblurring) real 3D restorative?

What kind of PSF is used? experimental or theoretical? If experimental, what
are the requirements? Must it be dyadic-sized? Must it be the same size as
the blurred image? Must it be an image with just a single microbead
recorded? Must the microbead be centered?

Finally, is there any User's Guide?


Thank you very much for your e-mails!


Dani

[Piotr Wendykier]

Dear Dani,

> I am a PhD student at the University of Barcelona, at the Department of
> Ecology. I am new in ImageJ, and I am interested in restoring 3D widefield
> images of bacteria and algae for intensity quantification purposes.
>
> I found the plugin Parallel Spectral Deconvolution (Piotr Wndykier) very
> interesting.

You will find answers to some of your questions here:

http://n2.nabble.com/Parallel-Spectral-Deconvolution-plugin-update-td635552.html#a635552


> I would need real restoration algorithms (constrained iterative or
> statistical image restoration, p.e, adn NOT no neighbour, closest neighbour
> or linear algorithms). Are the implemented algorithms (Generalized Tikhonov,
> Tikhonov and TSVD image deblurring) real 3D restorative?

The algorithms implemented in Parallel Spectral Deconvolution are fairly simple.
I have also implemented another plugin - Parallel Iterative
Deconvolution, available at

http://piotr.wendykier.googlepages.com/iterativedeconvolution

This plugin has four iterative algorithms to choose from and two of
them are nonnegatively
constrained. The description of the algorithms (links to the articles)
is on the plugin's website.


> What kind of PSF is used? experimental or theoretical? If experimental, what
> are the requirements? Must it be dyadic-sized? Must it be the same size as
> the blurred image? Must it be an image with just a single microbead
> recorded? Must the microbead be centered?

There are no limitations and Parallel Iterative Deconvolution supports
spatially variant PSFs.


> Finally, is there any User's Guide?

There is no user's guide, since there is no single recipe to do the
deconvolution.
You just need to try all the methods that I provide with different parameters
and hopefully one of them will work for you.

Regards,

Piotr Wendykier