Filtering (removing) single pixels

[Claude Bonnard]

Hello all,

Looking at the archive of the list, I coudn't find something which looks basic:

By analogy to the binary erosion filter, is there a possibility to remove only
single pixels (i.e. pixels surrounded by white) or in alternative pixels with
only one black neighbour?

Thanks for your advice!

Claude

--
Claude Bonnard Ph.D.
ISREC (Swiss Institute for Experimental Cancer Research)
Microscopy, Imaging and Morphology Core Facility [MIM]
Lab F216
Ch des Boveresses 155
CH-1066 Epalinges
Switzerland
email: [hidden email]
phone: [41-21]-692-5891
  fax: [41-21]-652-6933

[Filip Rooms]

Hi Claude,

> By analogy to the binary erosion filter, is there a possibility to remove only
> single pixels (i.e. pixels surrounded by white) or in alternative pixels with
> only one black neighbour?

If I understand you correctly, this sounds like a job for the median
filter... It takes all the pixel intensities in a window (say 3x3),
sorts them and takes then the median value. The effect is that
outliers in intensity (like so-called salt-and-pepper noise) are
removed (i.e., replaced by the median pixel value in the image)
without affecting the sharpness of edges. In ImageJ, it is called
"Despeckle" under "process", "noise" (if I can trust the analogy with
xv).

This works fine in case that just a small amount of pixels have been
corrupted (with a large value). However, it is no use to apply this
filter when all pixels are corrupted with, say, white Gaussian noise,
which means that all pixels are corrupted, but just by a small amount.

Hope this helps.

Kind regards,

Filip Rooms

[Joel Sheffield]

Depending on your purpose, try the median filter
(Process>Filters>Median).  See what it does to the "Enhance Me"
sample image.

Joel


Date sent:       Mon, 24 Apr 2006 14:56:03 +0000
Send reply to:   ImageJ Interest Group <[hidden email]>
From:           Claude Bonnard <[hidden email]>
Subject:         Filtering (removing) single pixels
To:             [hidden email]

> Hello all,
>
> Looking at the archive of the list, I coudn't find something which
> looks basic:
>
> By analogy to the binary erosion filter, is there a possibility to
> remove only single pixels (i.e. pixels surrounded by white) or in
> alternative pixels with only one black neighbour?
>
> Thanks for your advice!
>
> Claude
>
> --
> Claude Bonnard Ph.D.
> ISREC (Swiss Institute for Experimental Cancer Research)
> Microscopy, Imaging and Morphology Core Facility [MIM]
> Lab F216
> Ch des Boveresses 155
> CH-1066 Epalinges
> Switzerland
> email: [hidden email]
> phone: [41-21]-692-5891
>   fax: [41-21]-652-6933

Joel B. Sheffield, Ph.D
Department of Biology
Temple University
Philadelphia, PA 19122
Voice: 215 204 8839
e-mail: [hidden email]

[Paolo Ariano]

Claude Bonnard wrote:
> By analogy to the binary erosion filter, is there a possibility to remove only
> single pixels (i.e. pixels surrounded by white)

particle remover, does it.

--
Paolo Ariano
Neuroscience PhD @ UniTo

[Gabriel Landini]

> > By analogy to the binary erosion filter, is there a possibility to
> > remove only single pixels (i.e. pixels surrounded by white) or in
> > alternative pixels with only one black neighbour?

Yes, you can. Look at the "Count" option of Process>Binary.
You can select how many background pixels should be present before the
operation is applied.

G.

[Claude Bonnard]

Thanks all for the indications! This helped me solving my problem!

Regards,

Claude





On Apr 26,  2:26pm, Gabriel Landini wrote:

> Subject: Re: Filtering (removing) single pixels
> > > By analogy to the binary erosion filter, is there a possibility to
> > > remove only single pixels (i.e. pixels surrounded by white) or in
> > > alternative pixels with only one black neighbour?
>
> Yes, you can. Look at the "Count" option of Process>Binary.
> You can select how many background pixels should be present before the
> operation is applied.
>
> G.
>-- End of excerpt from Gabriel Landini

--
Claude Bonnard Ph.D.
ISREC (Swiss Institute for Experimental Cancer Research)
Microscopy, Imaging and Morphology Core Facility [MIM]
Lab F216
Ch des Boveresses 155
CH-1066 Epalinges
Switzerland
email: [hidden email]
phone: [41-21]-692-5891
  fax: [41-21]-652-6933