Re: [Matplotlib-users] applying an image's colormap to another image

[Fabrice S. <si...@lm...>]

Le mercredi 27 janvier 2016, Matteo Niccoli a écrit :
> Can something like this (which by the way I can't get to work):
> http://stackoverflow.com/questions/3114925/pil-convert-rgb-image-to-a
> -specific-8-bit-palette
> 
> What I would like to do is this:
> 1) Import an RGB image, which would have its own colormap - say this
> one for example:
> https://upload.wikimedia.org/wikipedia/commons/b/b3/Jupiter_new_hubble_view_above_pole.png

> 2) convert it to intensity, display the intensity color-mapped to the
> same colours the original RGB had.

According to the PNG header, this image does not have a palette (i.e. a
list of colors). The data chunks define the image as an array of NxMx3
values (N rows, M cols, 3 channels=no alpha), each value being defined
using 8 bits. I may however badly understand what you call the "own
colormap"...

You still can convert it to a grayscale img representing the intensity
(NxM values), but you then lose some information and you cannot display
it back with the same colors as originally. Because some different RGB
tuple are converted into the same intensity level, you can then not
discriminate them using the intensity image only.

Maybe there is some trick to convert to a grayscale image where those
RGB values are converted to almost-equal-but-different intensity levels
that would enable the later reconstruction, but I am not aware of...

Fabrice