Behavior of "Enhance Contrast" with "normalize" & "use stack histogram" options checked for 16-bit images

[lechristophe]

Hi,

From my understanding of the ImageJ manual, when using "Enhance Contrast"
on a 16-bit stack with the "normalize" and "use histogram" options checked,
it should adjust and scale all slices to the % saturation based on the
whole stack histogram, not the currently active slice. Is that right? It is
not what happens though, the output depends on the current active slice.

Consider the following macro code:
run("Bat Cochlea Renderings (449K)");
run("16-bit");
run("Add Slice");
run("Enhance Contrast...", "saturated=0.1 normalize process_all use");

As the adjustment is made starting from an all-black slice, everything on
the other slices get saturated.

By contrast, the same code but shifting the current slice to an non-empty
slice before calling "Enhance contrast" results in a different output:

run("Bat Cochlea Renderings (449K)");
run("16-bit");
run("Add Slice");
setSlice(1);
run("Enhance Contrast...", "saturated=0.1 normalize process_all use");

Is this the intended behavior or a bug? If it is, how can I adjust the
contrast of a whole stack to a given % saturation relative to the whole
stack histogram, and keeping the intensity relation between slices(like
when adjusting the contrast of confocal Z-stacks)?

Thanks,

Christophe

--
Christophe Leterrier
Researcher
Axonal Domains Architecture Team
CRN2M CNRS UMR 7286
Aix Marseille University, France

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

[Rasband, Wayne (NIH/NIMH) [E]]

On May 24, 2014, at 12:27 PM, Christophe Leterrier wrote:

> Hi,
>
> From my understanding of the ImageJ manual, when using "Enhance Contrast"
> on a 16-bit stack with the "normalize" and "use histogram" options checked,
> it should adjust and scale all slices to the % saturation based on the
> whole stack histogram, not the currently active slice. Is that right?

That is correct.

> It is
> not what happens though, the output depends on the current active slice.
>
> Consider the following macro code:
> run("Bat Cochlea Renderings (449K)");
> run("16-bit");
> run("Add Slice");
> run("Enhance Contrast...", "saturated=0.1 normalize process_all use");
>
> As the adjustment is made starting from an all-black slice, everything on
> the other slices get saturated.

Enhance Contrast sets the display range to 0-0 because the current slice has a pixel value range of 0-0. This causes the other slices to be saturated. Add

  setMinAndMax(0,65535);

to the end of the macro and the other slices will not be saturated.

> By contrast, the same code but shifting the current slice to an non-empty
> slice before calling "Enhance contrast" results in a different output:
>
> run("Bat Cochlea Renderings (449K)");
> run("16-bit");
> run("Add Slice");
> setSlice(1);
> run("Enhance Contrast...", "saturated=0.1 normalize process_all use");

The two macro generate the same output but they display it differently.

> Is this the intended behavior or a bug? If it is, how can I adjust the
> contrast of a whole stack to a given % saturation relative to the whole
> stack histogram, and keeping the intensity relation between slices(like
> when adjusting the contrast of confocal Z-stacks)?

It the latest daily build (1.49b10), Enhance Contrast sets the display range of 16-bit images to 0-65536 when "Normalize" is checked.

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