Converting RGB Image to Grayscale Intensity on arbitrary Colormap

We have some thermal videos that have a colorscale presented along the side. I would like to run some thermal analysis on video frame regions. If I extract the colorscale from the video, how could I apply that to the image such that the result would be grayscale based on the intensity of the colorscale?

I have obtained a cmap for the custom scale by reading in that area, taking the mean over each column, resulting in a 1601x3 double, but I don't know how to map that onto a grayscale image as intensity values.

Answer 1

[Image Analyst]

People ask this all the time. So much so that I have a canned demo for it. See attached.

Adapt as needed, like to specify where in the image your image and colorbar are located.

[Matt Gaidica]

Excellent, thanks again!

Answer 2

[DGM]

Something like this might be a start:

% you already found a colormap
cmap = parula(64);

% generate false-color test image
A = imread('cameraman.tif');
Aind = gray2ind(A,size(cmap,1)); % uniform quantization
Afalse = ind2rgb(Aind,cmap); % apply colormap
imshow(Afalse)

% convert back to intensity image
Arecovered = double(rgb2ind(Afalse,cmap))/(size(cmap,1)-1);
figure; imshow(Arecovered)

immse(im2double(A),Arecovered) % not perfect, but pretty good

ans = 2.9731e-05

[Matt Gaidica]

I think this is along the right track, but my input image is RGB. It's throwing an error in ind2rgb()on imresize(). It's not immediate to me how to adjust it. Thanks DGM

[DGM]

All you need is the second part. The first part is just generating a falsecolor test image, in concept like the one you already have. Your image would be Afalse, and your colormap should be cmap.

[Matt Gaidica]

Thanks for your time DGM!