Color detection and display its name

Color detection project

• take colors in circles or squares
• print it
• take a photo of this print
• then put it in your code for color detection
• It should be able to display the name of that color

I have started this project with very simple concept of masking. But colors exists in hundreds or thousands of shades. In my project code should be able to detect any shade and any color. I am thinking to use the concept of Euclidean distance to find the distance of color from origin, but i can not really do it.

If someone could help me please post your solution.

Here is the code i used. it detects only four colors but from one image only.

%------------------------------------
% Read the image into an array.
filename=input('Input the filename of the image>');
RGB = imread(filename);
RGB = imread('D:\College\Project\Project\Cube.png');
%------------------------------------
% Display the original image.
subplot(2, 3, 1);
imshow(RGB);
title('Original RGB Image');

yellowMask = colorDetectHSV(RGB, [0.2 0.9 0], [0.05 0.05 0.05]);
redMask =colorDetectHSV(RGB, [1 1 1], [0.1 0.2 0.2]);
blueMask =colorDetectHSV(RGB, [0.6 0.5 0.6], [0.2 0.3 0.4]);
% Or can try this for blue color:
% blueMask =colorDetectHSV(RGB, [1 0.5 0], [0.4 0.4 0.3]);
greenMask =colorDetectHSV(RGB, [0.369 0.786 0.769], [0.2 0.2 0.2]);

% Display them.
subplot(2, 3, 2);
imshow(yellowMask);
title('Yellow Mask');
subplot(2, 3, 3);
imshow(greenMask);
title('Green Mask');
subplot(2, 3, 4);
imshow(blueMask);
title('Blue Mask');
subplot(2, 3, 5);
imshow(redMask);
title('Red Mask');

[Christian Lenz]

I don't quite well understand your question... Every color has allready a specific value in your RGB-image. Try:

 RGBimg = imread( ... );
 RGBimg = double( RGBimg );
 Rimg = RGBimg ./ max(max(max( RGBimg ) ) );

[SARBE]

actually this code is able to detect only four colors. i don't know how to expand it so that it can detect a number of colors.

[Walter Roberson]

Did you look at the work of John D'Errico, or of Image Analyst ?

Answer 1

[Walter Roberson]

Euclidean distance is not enough. Look in the File Exchange for John D'Errico's Fuzzy Color Detection.

But if your task really is to "detect any shade and any color", you should consider giving up before you start any development. Seriously. What you want to do cannot reliably be done. There is no scientific definition for what each color is, or where one shade ends and another begins. And color perception is personal, not absolute. See http://blog.xkcd.com/2010/05/03/color-survey-results/

Answer 2

[Image Analyst]

I have several color detection applications in my File Exchange. http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862 Each uses a different method. Feel free to adapt any of them.

[Navodita Das]

Above file exchanges are very useful, thank you for that.

Is it possible to set and reference axes in image and find the color point's coordinates axis w.r.t. reference axes.

like we enter an image-

and matlab will set the reference axes on its own

then

will get the coordinates of color points w.r.t the ref. axis.

to get (x,y) of green,blue.

is it possible to get?

[Image Analyst]

Sure, just subtract the x and y coordinates of the origin from any other x and y coordinates you get. Be careful not to get confused with thinking (x,y) is how you index an image. Images are matrices and are indexed (row, column) which is (y, x), NOT (x, y).

[Navodita Das]

Ok, thank you.

[Navodita Das]

Suppose, an image contains red point in it. Can I set that red point as origin of coordinate system?

[Image Analyst]

Yes, of course.