Lab vs RGB vs HSV color spaces for color detection?

I'm having trouble understanding why there exist multiple color spaces. That is, informatically, all three spaces have the same color information content, so why would one perform better or worse for selecting out, say, red?

I saw from the example demo by ImageAnalyst that lab space is very good for detecting a particular color selected by the user, but I just don't get why the performance should be different for any color space? does it have to do with some computational issues?

thank you for your time.

Answer 1

[Image Analyst]

You have to look at the shape of the gamut in the different color spaces to understand which color space will be best. Unfortunately MATLAB does not have a gamut visualizer like imageJ does ( http://rsb.info.nih.gov/ij/plugins/color-inspector.html) so you'll have to use that to understand your gamut. I've been telling the Mathworks for years that it's my number one most requested upgrade they can do. For example imagine that you have a gamut that is like a cigar shape extending out into the orange region of LAB/HSV space. You're not going to be able to carve out those pixels with a box, which is what you get if you set thresholds in LAB color space, but if you set thresholds on H, you can carve out a sector going radially, for example you can get all hues going from 40 to 50 degrees, which is the orange region.

[Image Analyst]

Code for a shortcut to the imageJ gamut inspector:

% Windows installation instructions:
% First, download and install ImageJ from here:
% http://rsb.info.nih.gov/ij/download.html
% Then download the 3D Color Inspector plug-in for ImageJ here:
% http://rsb.info.nih.gov/ij/plugins/color-inspector.html
% and save it here:
% C:/Program Files/ImageJ/plugins/Color
%
% Now we need to enable the plug-in to work with MATLAB.
% First open your classpath file:
% C:\Program Files\MATLAB\R2013a\toolbox\local\classpath.txt
% for editing.  Then add the following lines as new lines to your classpath file:
% C:/Program Files/ImageJ/ij.jar
% C:\Program Files (x86)\ImageJ\plugins\Color\Color_Inspector_3D.jar

clc;  % Clear the MATLAB command window.

jarFileName = 'C:\Program Files (x86)\ImageJ\plugins\Color\Color_Inspector_3D.jar';
if ~exist(jarFileName, 'file')
  warningMessage = sprintf('Error: Color Inspector jar file does not exist:\n%s', jarFileName);
  warndlg(warningMessage);
  return;
end
% User must browse for the file they want to open.
% Set up a starting/initial folder that they will start browsing from.
startingFolder = pwd; % Change to something specific if you want.
if ~exist(startingFolder, 'dir')
  startingFolder = pwd;
end
% Bring up the "Open File" dialog box.
[baseFileName, folder] = uigetfile({'*.jpg;*.tif;*.png;*.gif;*.bmp','Image Files (jpg,tif,png,bmp,gif)';...
'*.*','All Files (*.*)' },'Specify image file',  startingFolder);

if baseFileName ~= 0
  % If they didn't click Cancel, build the full filename:
  fullFileName = fullfile(folder, baseFileName)
  % Now pass that into the 3D Color Inspector plugin, and you're done.
  color_inspector(fullFileName)
end

[Image Analyst]

My MATLAB code for a color gamut visualizer, though it's primitive compared to the ImageJ plug-in.

% Creates a 3D scatter plot of an RGB color gamut of a color image.

clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
imtool close all;  % Close all imtool figures.
clear;  % Erase all existing variables.
workspace;  % Make sure the workspace panel is showing.
format longg;
format compact;
fontSize = 20;

% Read in a standard MATLAB color demo image.
folder = fullfile(matlabroot, '\toolbox\images\imdemos');
baseFileName = 'peppers.png';
% Get the full filename, with path prepended.
fullFileName = fullfile(folder, baseFileName);
if ~exist(fullFileName, 'file')
  % Didn't find it there.  Check the search path for it.
  fullFileName = baseFileName; % No path this time.
  if ~exist(fullFileName, 'file')
    % Still didn't find it.  Alert user.
    errorMessage = sprintf('Error: %s does not exist.', fullFileName);
    uiwait(warndlg(errorMessage));
    return;
  end
end
rgbImage = imread(fullFileName);
% Get the dimensions of the image.  numberOfColorBands should be = 3.
[rows columns numberOfColorBands] = size(rgbImage);
% Display the original color image.
imshow(rgbImage);
title('Original Color Image', 'FontSize', fontSize);
% Enlarge figure to full screen.
set(gcf, 'units','normalized','outerposition',[0 0 1 1]);

% Extract the individual red, green, and blue color channels.
redChannel = rgbImage(:, :, 1);
greenChannel = rgbImage(:, :, 2);
blueChannel = rgbImage(:, :, 3);

% Construct the 3D color gamut.
gamut3D = zeros(256,256,256);
for column = 1: columns
  for row = 1 : rows
    rIndex = redChannel(row, column) + 1;
    gIndex = greenChannel(row, column) + 1;
    bIndex = blueChannel(row, column) + 1;
    gamut3D(rIndex, gIndex, bIndex) = gamut3D(rIndex, gIndex, bIndex) + 1;
  end
end

% Get a list of non-zero colors so we can put it into scatter3()
% so that we can visualize the colors that are present.
r = zeros(256, 1);
g = zeros(256, 1);
b = zeros(256, 1);
nonZeroPixel = 1;
for red = 1 : 256
  for green = 1: 256
    for blue = 1: 256
      if (gamut3D(red, green, blue) > 1)
        % Record the RGB position of the color.
        r(nonZeroPixel) = red;
        g(nonZeroPixel) = green;
        b(nonZeroPixel) = blue;
        nonZeroPixel = nonZeroPixel + 1;
      end
    end
  end
end
figure;
scatter3(r, g, b, 3);
xlabel('R', 'FontSize', fontSize);
ylabel('G', 'FontSize', fontSize);
zlabel('B', 'FontSize', fontSize);
msgbox('Click the rotation icon to change your point of view.');

[David]

ohhhh...let me recap for my clarity.

I have an image, it's color can be represented in some color space, such as HSV, LAB, RGB. The degree to which each color is present in my image will dictate the "shape" it takes in selected color space.

My color selection algorithm (i.e. thresholding as do the delta thing) is basically carving out a box.

So, depending on how box-like my selected color is in the given space of choice, will determine how seperable it will be.

is that all correct? Either way, that gamut function you describe seems far more essential than a lot of the newer tools they have!

[Image Analyst]

Yeah, that's basically it. But it depends on how you look at it. If your gamut is a pie-shaped sector, then it makes sense to use HSV space, where thresholding in hsv space is like carving out a box in hsv space, but I prefer to think of it as carving out a pie-shaped sector in the 3D lab/hsv space.

If you have skin color, then you have a boomerang shaped gamut and no simple thresholding in any color space will get it. You can get a sector by using hue angle to carve out a pie shape - fine that gets you the hue - but doing just that will get you gray things too (like lab=50,0,0) and we know there are no gray people (unless they're dead or zombies), so you need to do a little more to carve out the inner elbow of the boomerang to exclude gray objects.

[David]

that is very helpful...I think I'm getting it. I'll explore more. thank you very much for your input.

Follower of the gamut() movement.

[Image Analyst]

Keep it up. It takes a while of seeing it over and over again and each time you see it, a little more sinks it. But it's not the kind of thing you see once and have mastered. I'm in the process of returning from a literally around-the-world trip (yes, flew east the whole time) where I taught well over a hundred people in 6 countries about color science. I'd estimate they retain about 10% of what I tell them. Next month I'm training in South America.

In addition to my File Exchange, take a look at Stephen Westland's: http://www.mathworks.com/matlabcentral/fileexchange/40640-computational-colour-science-using-matlab-2e. He's one of a few real experts in color science, but even he'll admit, like me, that there are just some things we don't know yet.

[Sabanam]

@Image Analyst.

Its really helpful to me also..

[Image Analyst]

Glad I could help. You can also vote for my answer if you want.

[Walter Roberson]

mohammad moosazadeh comments to Image Analyst about https://www.mathworks.com/matlabcentral/answers/84068-lab-vs-rgb-vs-hsv-color-spaces-for-color-detection#comment_163055

its useful