Converting 3D Image Into Colormap

Question

I have a 3D image I want to use as a colormap. I have real-time data streaming in and depending on my current position(s), I want my graph to display a certain color based on my 3D image.

I did this earlier with success using an image of size 1080x1920 (2D image) using the following code to convert it to an image.

Color=double(imresize(RGBparam.image,[81 144]))/255;

then plotting it by setting my data (I already created figure earlier...that code shouldn't matter) where size(myData(:,:1)) outputs 81x144...the Ydata and Zdata also have the same dimensions.

set(Fig.figim, 'CData', Color);
set(Fig.figim, 'XData', myData(:,:,1) + positionX)
set(Fig.figim, 'YData', myData(:,:,2) + positionY);
set(Fig.figim, 'ZData', myData(:,:,3) + positionZ);

So that worked fine! Now I have a new 3D image. The image has dimensions 512x512x351. I tried doing the same code but it does not work. I tried this:

Color=double(imresize(DICOMparam.Vd,[81 144]))/255;

Then I try to plot like this

set(Fig.figim, 'CData', Color);
set(Fig.figim, 'XData', myData(:,:,1) + positionX)
set(Fig.figim, 'YData', myData(:,:,2) + positionY);
set(Fig.figim, 'ZData', myData(:,:,3) + positionZ);

But then I get the error: "Color must be an M-by-N matrix or an M-by-N-by-3 array"

Now I tried changing things so it turns out to be an M-by-N-by-3 array, but first of all, I don't want this...I want to keep more than just 3 samples of the z-dimension of my color map....and second of all...it spit out a bunch of errors at me anyways. So in conclusion, I simply want this 3D colormap to work when I plot my data, which has x, y, and z components. Thanks for the help, let me know if anything is unclear!

Answer 1

Imagine your 2D/3D image as a square/cube of color. What you want to do (if I understand correctly), is color each point in myData depending on where it falls into that square/cube.

However, that's not how 'CData' works in Matlab. It maps colours based on indexed location, not the actual values in the matrix. Therefore if your data is M-by-N, your colormap is M-by-N - e.g. regardless of the actual values of the data, myData(1,1) is coloured according to myColormap(1,1).

In the case of the M-by-N-by-3 array, these are still just indexed locations, but the three values in each case define the colour in RGB.

---

It is possible to do what you want, you just have to tell Matlab directly what color in that 512 x 512 x 351 cube you want to apply to each point. Presuming you have some way of converting the x, y, z positions in myData into approximate coordinates in the 512 x 512 x 351 colorspace:

xData = myData(:,:,1); 
yData = myData(:,:,2);
zData = myData(:,:,3);

% some magic here happens to make xData 1-512, yData 1-512, zData 1-351
% may require shifts, scaling, rounding as appropriate
% depends on what your myData contains exactly
% values should be integers

% now we find the positions of those points within the 512 x 512 x 351 colorspace
% size of newcmap should be 81 x 144 in your case
cind = sub2ind(size(Color),xData,yData,zData);
newcmap = Color(cind);

set(Fig.figim, 'CData', newcmap);