# Applying a color for each pair of spherical angles through the hue and lightness parameters

7 views (last 30 days)
Richard Wood on 17 Oct 2023
Commented: DGM on 29 Nov 2023
Hello eveyone
I have characterized which is the the polar and azimuthal angles of a vector representation in each point of a 2D plane. Given that the modulus of each vector is constant thorugh all the system, to plot the information, I would like to use some kind of spherical color representation like this
with the hue represented with some color palette similar to the second or third ones depicted by @DGM in this post How can I plot circular colormap for 0-24 hour phase? Ideally, I would need also that depending on the polar angle, the lightness change from white to black, like in the image above. I have done in the past a question in the same like Creation of a 2D colormap to represent simultaneously information about the polar and azimuthal angles, but I still haven't been able to find a way to do it. Any ideas?
##### 2 CommentsShow NoneHide None
Image Analyst on 17 Oct 2023
It's not clear to me exactly what you want -- a gamut visualization, an image of yours with some colormap applied, an Nx3 colormap matrix, or something else. So, I have no ideas other than those that may possibly be what you want. Maybe youi want colorcloud - who knows?
DGM on 17 Oct 2023
Edited: DGM on 17 Oct 2023
This could be done in HSL or HSY,but HSV won't have the symmetry. Visualizing the map as a 3d solid is not very readable imo.
I can't really work on this right now, since I'm not at the computer.

DGM on 18 Oct 2023
Edited: DGM on 18 Oct 2023
Rereading the question, I take it that the 3D representation is just an example, not necessarily a goal. Given that the vector magnitudes are constant, this can be reduced to a 2D map.
I'm going to do this example using MIMT tools, since MATLAB doesn't have tools for any of these models other than HSV.
% setup
azrange = [0 2*pi];
elrange = [-pi/2 pi/2];
colormodel = 'huslp'; % 'hsv','hsl','hsyp','huslp'
% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% constructing some fake az,el data
sz = [500 700];
rng(44524) % just for demo consistency with random test data
AZ = imresize(perlin(round(sz/10)),sz); % or low-freq random noise
AZ = imrescale(AZ,imrange(AZ),azrange);
EL = imresize(perlin(round(sz/10)),sz);
EL = imrescale(EL,imrange(EL),elrange);
% constructing a color image representing the az,el data
% assuming all angles have been wrapped to the appropriate intervals
H = imrescale(AZ,azrange,[0 360]); % this could also be done with rescale()
L = imrescale(EL,elrange,[0 1]);
S = ones(sz); % 100% saturation
% assemble the RGB image using some normalized color model
switch lower(colormodel)
case 'hsv'
azelpict = hsv2rgb(cat(3,H/360,S,L)); % base MATLAB
case 'hsl'
azelpict = hsl2rgb(cat(3,H,S,L)); % MIMT only (or other FEX tools)
case 'hsyp'
azelpict = hsy2rgb(cat(3,H,S,L),'pastel'); % MIMT only
case 'huslp'
azelpict = husl2rgb(cat(3,H,S*100,L*100),'labp'); % MIMT only
end
% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% constructing a reference image to replace the colorbar
N = 256;
cbaz = linspace(azrange(1),azrange(2),N);
cbel = linspace(elrange(1),elrange(2),N);
[cbaz cbel] = meshgrid(cbaz,cbel);
% assuming all angles have been wrapped to the appropriate intervals
H = imrescale(cbaz,azrange,[0 360]);
L = imrescale(cbel,elrange,[0 1]);
S = ones(size(cbaz)); % 100% saturation
% assemble the RGB image
switch lower(colormodel)
case 'hsv'
cbpict = hsv2rgb(cat(3,H/360,S,L)); % base MATLAB
case 'hsl'
cbpict = hsl2rgb(cat(3,H,S,L)); % MIMT only (or other FEX tools)
case 'hsyp'
cbpict = hsy2rgb(cat(3,H,S,L),'pastel'); % MIMT only
case 'huslp'
cbpict = husl2rgb(cat(3,H,S*100,L*100),'labp'); % MIMT only
end
% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% display the things
subplot(1,2,1)
image(azelpict)
axis('equal','tight')
title('My Data')
subplot(1,2,2)
image(cbpict,'xdata',azrange/pi,'ydata',elrange/pi)
axis('square','tight')
set(gca,'ydir','normal')
These are the figures showing the same random data represented using HSV, HSL, HSYp, and HuSLpab.
It's up to you to decide what you think is most readable in your application. The poor uniformity of HSV/HSL leads to a strong perceived banding effect in smooth data. HuSLp inherits the uniformity of LAB, but its limited chroma range leads to reduced distinctness. This tradeoff is similar to part of the argument often used in favor of rainbow colormaps in 1D colormapping contexts (e.g. jet() vs parula()). HSYp is something in-between (chroma-normalized YPbPr). It has more chroma range than HuSLpab. Nobody would claim it to be uniform, but it's more so than HSV/HSL.
If, for the purpose of readability, it's desired to discretize/quantize the mapping process (i.e. so that there are fewer distinct colors in the 2D map), that may also be an option. I'll leave that for further discussion if needed.
This answer may also work (using a pregenerated arbitrary 2D colormap)
Tangentially related:
Image Analyst on 29 Nov 2023
@Richard Wood in addition to what Walter just said (link to Perlin noise), you might try my attached demo, based on Prof. Peter Kovesi's noiseonf.m code, that makes cloud-like images with 1/f noise.
DGM on 29 Nov 2023
Walter already linked to the tools I used. The use of perlin() was just for demonstration. It was a cheap way to make pseudodata that was formed of cloudlike blobs as in your other posts. If instead we used overlaid radial gradients, it might highlight different aspects of the differences between the color models.
% ...
% ...
For smooth colormaps:
For quantized maps:
... and yes, radgrad() is also from MIMT.

### Categories

Find more on Red in Help Center and File Exchange

### Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!