Re: color calibration of microscopic images

From: Beatrice (pappagallo16_at_libero.it)
Date: 09/08/04 

Date: 8 Sep 2004 06:52:35 -0700

Hi,

thank you very very much for all your comments. You all make very good
points. I also have some comments myself.

Beatrice

David, your first assumptions are correct.

> I would have thought filter paper may be a bit variable. Can you use a
> Macbeth calibration white standard or something?

What do you mean by variable? You mean that the paper structure would
be seen under the microscope?
I am using this filter paper after reading on "Color science in the
examination of museum objects" by Feller that a stack of Millipore
filters with a pore size of 0.025 um can be recommended as white
reference standard for spectrophotometry. I thought that I could use
it for what I am doing as well. I don't know what is the pore size of
the paper that I am using, but it definitively looks far more uniform
than the white patch of the Macbeth ColorChecker chart.

> You could try defocussing the microscope to blur the image. This should
> not change the colour balance.

I have tried doing this (keeping an eye on the live histogram, in
order to check that its peak is only narrowing but not shifting), but
I feel very uncomfortable in doing this, since the level of defocus is
very difficult to control and therefore seems to me very arbitrary.
However, results don't seem to change.

> Do you find a colour shift as you change the exposure time or ISO
> setting on the camera?

Besides a change in brightness? I would not know for sure. But I can
tell you that I have acquired images of the grey patches of the Gretag
Macbeth chart, at different exposure times, and I see a nonlinear
effect (even before the histogram becomes saturated).

> If so, do neutral density
> filters change the colour?

White point does not change. But I think that I would have the same
effect as of the previous point, because of the nonlinear response of
the camera.

> Failing these, try
> using a pair of polarisers in a mount in which one can be rotated
> relative to the other to give varied extinction (snip)

Same problem again. I wouldn't use polarizers anyway, one reason being
that some pigment materials are birefringent.

> I don't know what your patches are, in detail (snip).

My samples are transversal sections, so what I see is the whole layer
build-up, possibly from the first paint layer above the painting
support to the varnish.

> If I were doing this I would want to compare with some modern paint
> samples as a control.

Making my own target with paints. This is probably the best solution.
The optical properties would also be much more similar.

> BTW, I still think a spectrophotometer would give a much more rigorous
> definition of spectral properties (snip).

Certainly. But the aim of this work is not to identify pigment
materials. I simply want to quantify the change of appearance of a
paint cross-section, or to measure the difference of appearance of
different cross-sections of their light-microscopic images. Of course,
I might also want to compare not color images but other properties
(mass spectra, visible-light spectra, etc), but this will come
afterwards.

Aaron:

> (snip) it is probably the case that your samples
> will not be homgeneous nor will they be consistant over different
> portions of the canvas.

Yes. This is the reason why I want to make these comparisons.

> the method of abrasion and the
> abrasive itself will affect the surface texture and may also leave a
> trace residue embedded in the specimen surface which may also affect
> color..

This is in fact a real problem for us, and we could not find a
solution.
But in my case is maybe not critical, as I mostly want to compare
colors (read also my reply to David).
Certainly I would like to have a good reproduction of colors, but I
don't need much accuracy. Please don't think that I am not looking for
absolute color, nor I am assuming there is. My studies in physics help
me in this sense.
The main reason why I want to do color calibration is that in this way
I would be able to compare images acquired in different conditions and
in different sessions. What I mean is not to get an absolute color,
but to compensate for instrumental effects and to convert my images
acquired in different conditions to one only (color temperature of the
light source, neutral filters, exposure time, etc).
Just I am not sure whether doing a color calibration is sufficient or
not, and whether I am applying it correctly...

...

I am aware of many of many of the other things you say (thanks for
pointing them out!!), and that there are many variables. The best
thing to do is to determine the ones that are crucial for my specific
aim. I won't need to take all of them into account to have a
meaningful result.

Human perception is an interesting issue. Do I want to compare my
images in terms of their physical properties only or do I also want to
account for human perception? I think both approaches make sense,
although in general I would expect different results.
And if take into account human perception, what are the relevant
aspects for a meaningful result?

Oh, and thanks for the link. I will read it. And even if it's basic,
there might still be things I was not aware of.

Relevant Pages

  • Re: color calibration of microscopic images
    ... >filters with a pore size of 0.025 um can be recommended as white ... >tell you that I have acquired images of the grey patches of the Gretag ... >> If I were doing this I would want to compare with some modern paint ... a spectrophotometer would at least allow you ...
    (sci.techniques.microscopy)
  • Re: The sickening reality of high ISO on a P&S
    ... Go to any camera review and you'll see all the support you need. ... There are cases where the FZ8 ... There are no images to see... ... I can't provide images to compare because I wasn't at those particular ...
    (rec.photo.digital)
  • Re: Compare files?
    ... a file compare utility to claim them as different. ... images that look similar. ... And the software it DOES list is NOT the file I have, so beware anything should you decide to download from there. ...
    (microsoft.public.windowsxp.help_and_support)
  • Re: 35mm film scanner questions
    ... I have found that whenever there is an attempt to directly compare digital ... and analog images, there is always something that renders the comparison ... don't know of any digital cameras that come even close to that...what can we ... the resolving ability of film and film camera lens has ...
    (rec.photo.digital)
  • Re: D80 noise versus Rebel XTi noise at 1600 ISO
    ... Rich didn't have any conclusions, ... data, nor any equalized testing criteria, and the crops could be ... false data and still has false images and claims on his site). ... Where's the exif and original image so we can compare? ...
    (rec.photo.digital.slr-systems)
Quantcast