Posted by
Peter Haub on
Oct 31, 2014; 4:08pm
URL: http://imagej.273.s1.nabble.com/CIELab-Color-Averaging-tp5010215p5010269.html
Hallo Albatross,
have you thought about treating your images as absorbance images? Maybe
it is worth to try since they look like brightfield reflectance images -
means that the color is probably created by the spectral absorbance of
the object.
If you calculate the values A = -log((I-B)/(Io-B) for the three channels
RGB you can measure the mean absorbance vector for each object. This
vectors should describe the object's absorbance characteristics (-
sometimes called color :-).
I is the intensity of the RGB channel images.
B is a background values (not image but noise background)
I is the maximum intensity in your image (value of a white object)
I have done that by assuming B=10, Io=255.
I calculated the average vector from the absorbance vectors of the
objects. The angle between the vectors and the average are:
Object I (Upper Left): 0.0817°
Object II (Upper Right): 0.1187°
Object III (Lower Left): 0.0855°
Object IV (Lower Right): 0.1433°
I think this results nicely describe the scene (the two left object have
a similar color and are closer to an estimated mean color).
Note, this Lambert-Beer-like approach is very much simplified and is not
really describing physical reality. In any case you have to consider
that the illumination, imaging parameters etc. have to be stable and
unchanged between measurements. The calculation should be performed in
32bit images. If a back-mapping to 8bit is necessary, e.g. to visualize
the combined absorbance values in the 3D Color Inspector, it should be
performed with a unified maximum display value for all 3 channels.
Maybe you will give it a try.
Regards,
Peter
On 31.10.2014 00:33, Albatross wrote:
> Thank you for your response, great info.
>
> And some more questions:
>
> 1) *How can I determine if my image is in the sRGB format?* I am using
> images from a scanner, the metadata from the image only says that it is an
> RGB image.
>
> 2) *So when the 'Average Color' function is using CIELab averaging, is it
> converting RGB values into the CIELab color space?*
>
> 3) *Is there a better way to get a mean color value for R, G, and B per a
> ROI (see photo example below)? Can I batch process this?* I've tried using
> the multi-measure tool (after splitting the image into R, G, B slices) in
> the ROI manager, but the data output puts the values in an order that is not
> very useful. To be more specific, the time it would take me to reformat the
> multi-measure output in excel would be too great. I need the output to be
> in order by R, G, B per ROI per image so that I can convert the values to H,
> S, B.
>
> Any help would be greatly appreciated, thank you!
>
> <
http://imagej.1557.x6.nabble.com/file/n5010257/20140204_Plastic_Photo_Example_B.jpg>
>
>
>
> --
> View this message in context:
http://imagej.1557.x6.nabble.com/CIELab-Color-Averaging-tp5010215p5010257.html> Sent from the ImageJ mailing list archive at Nabble.com.
>
> --
> ImageJ mailing list:
http://imagej.nih.gov/ij/list.html>
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html