Re: Reduce contiguous area in Find Maxima plugin

[Michael Schmid]

Hi Guiseppe,

please send questions about ImageJ to the mailing list, not to me personally; there might be many others having the same wishes or problems!

In principle, one can reduce the 'noise tolerance' with distance. Note, however, that this can be done accurately only if the maximum is a single point.  The result will become unpredictable if there are several pixels having exactly the value of the maximum (an arbitrary one of these will be selected; it would be a hell of work to do it in a well-defined way).
What would be a good function for decreasing the noise tolerance with distance? A slow decrease line 1/(1 + r/r0)?  A function that approaches zero rather rapidly, like a Gaussian?

I also have Philippe's recent whishlist in mind, see
https://list.nih.gov/cgi-bin/wa.exe?A2=ind1406&L=IMAGEJ&F=&S=&P=177817

What do you think about the following idea:

Having an external "Find Maxima Plus" plugin, with the following features:

- Philippe's bells&whistles, see the posting above,
- Threshold being a combination of a fixed value and a fraction (percentage of the peak value)
- Maybe minimum distance between two maxima?
- Some reduction of the noise tolerance with distance, see above.

I think that this probably should not go into standard ImageJ, it would be too complex for most users. It would take a while, but I could try doing it this summer.

Regards,

Michael
________________________________________________________________
On Jun 30, 2014, at 04:38, Giuseppe Lucarelli wrote:

> Dear Michael,
>
> I am using the Find Maxima plugin in imagej but in ROI with visible (by eye) >1 maxima the plugin only finds one unless I lower the noise tolerance to a value that picks up other undesired points in the ROI.
> The problem is that the two peaks I need to select, although far apart, are not separated by low value points ("valley") which affects the plugin.
> I thought I could modify the plugin so that the valley between two maxima does not need to be so low.
>
> What are your thoughts on this?
> If you think I should modify the plugin, how do I do that?
>
> Thanks a lot for your help.
>
> Regards,
> Giuseppe Lucarelli

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Giuseppe Lucarelli]

Thanks a lot for your response Michael!

I looked at the source code of the plugin again and realised that the
cleanupMaxima
function is what dictates the minimum distance between two or more maxima.
Is this correct?

Thanks a lot


On 2 July 2014 02:30, Michael Schmid <[hidden email]> wrote:

> Hi Guiseppe,
>
> please send questions about ImageJ to the mailing list, not to me
> personally; there might be many others having the same wishes or problems!
>
> In principle, one can reduce the 'noise tolerance' with distance. Note,
> however, that this can be done accurately only if the maximum is a single
> point.  The result will become unpredictable if there are several pixels
> having exactly the value of the maximum (an arbitrary one of these will be
> selected; it would be a hell of work to do it in a well-defined way).
> What would be a good function for decreasing the noise tolerance with
> distance? A slow decrease line 1/(1 + r/r0)?  A function that approaches
> zero rather rapidly, like a Gaussian?
>
> I also have Philippe's recent whishlist in mind, see
> https://list.nih.gov/cgi-bin/wa.exe?A2=ind1406&L=IMAGEJ&F=&S=&P=177817
>
> What do you think about the following idea:
>
> Having an external "Find Maxima Plus" plugin, with the following features:
>
> - Philippe's bells&whistles, see the posting above,
> - Threshold being a combination of a fixed value and a fraction
> (percentage of the peak value)
> - Maybe minimum distance between two maxima?
> - Some reduction of the noise tolerance with distance, see above.
>
> I think that this probably should not go into standard ImageJ, it would be
> too complex for most users. It would take a while, but I could try doing it
> this summer.
>
> Regards,
>
> Michael
> ________________________________________________________________
> On Jun 30, 2014, at 04:38, Giuseppe Lucarelli wrote:
>
> > Dear Michael,
> >
> > I am using the Find Maxima plugin in imagej but in ROI with visible (by
> eye) >1 maxima the plugin only finds one unless I lower the noise tolerance
> to a value that picks up other undesired points in the ROI.
> > The problem is that the two peaks I need to select, although far apart,
> are not separated by low value points ("valley") which affects the plugin.
> > I thought I could modify the plugin so that the valley between two
> maxima does not need to be so low.
> >
> > What are your thoughts on this?
> > If you think I should modify the plugin, how do I do that?
> >
> > Thanks a lot for your help.
> >
> > Regards,
> > Giuseppe Lucarelli
>
>

--
Giuseppe Lucarelli
Biochemistry PhD student
Prescott/Devenish Lab, Bld 13 D224a
PH  990 53785

<[hidden email]>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html