Distance of rgb pixel from each pixel of image

I am using impixel to select the similar pixel values of rgb image and find their mean. those mean rgb values are Ra,Ba,Ga. I want to find the eucledian distance of each rgb valued pixel from these mean values but i am unable to do so.

im = imread('coloredChips.png');

p=impixel(im)

R=p(:,1); %extracting red value in rgb matrix

G=p(:,2); %extracting green value in rgb matrix

B=p(:,3); %extracting blue value in rgb matrix

Ra=mean(R);

Ga=mean(G);

Ba=mean(B);

r1=im(:,:,1); %separating red, green,blue layer of image

g1=im(:,:,2);

b1=im(:,:,3);

rr=(r1-Ra).^2;

gg=(g1-Ga).^2;

bb=(b1-Ba).^2;

v=sqrt(rr+gg+bb);

Answer 1

[Image Analyst]

impixel just gives you the value at a single point. So the mean there is just going to be the value there.

If you want the RGB as separate color channels use imsplit

rgbImage = imread('peppers.png');
imshow(rgbImage);
axis('on', 'image');
[R, G, B] = imsplit(rgbImage);
uiwait(helpdlg('Select a point'))
[x,y] = ginput(1) % Ask user for a point.
x = round(x); % Convert to integer
y = round(y);
hold on;
plot(x, y, 'r+', 'LineWidth', 2, 'MarkerSize', 200)
rPoint = double(R(y, x));
gPoint = double(G(y, x));
bPoint = double(B(y, x));
% Get mean of whole image
meanR = mean2(R)
meanG = mean2(G)
meanB = mean2(B)
rr=(rPoint-meanR).^2;
gg=(gPoint-meanG).^2;
bb=(bPoint-meanB).^2;
v=sqrt(rr+gg+bb)

Also see my attached "magic wand" demo to get all colors within a certain delta E (color distance) of a region.

[Maleeha Syed]

Hi!

i actually want to obtain the distance of every pixel of the image from the mean R G B values not of a single point from the mean values if you could guide me in that part of the code

[DGM]

When you say "the mean RGB values" what exactly do you mean? Do you want the mean from the entire image? Or do you want the local mean in a region around the selected location?

[Maleeha Syed]

using impixel i select certain locations of the image i.e., for instance i click on red color at different locations and then i would get a vector with the rgb values at every pixel as shown in figure and then i separate them out and get the mean and i get mean of r value , g value , b value i.e.Ra,Ga,Ba . I want to now find the distance of this mean Ra,Ga,Ba value from every rgb valued pixel in the image using a loop i.e, distance of evry pixel is sqrt((r-Ra)^2+(g-Ga)^2+(b-Ba)^2)

[Image Analyst]

Try this:

rgbImage = imread('peppers.png');
subplot(2, 1, 1);
imshow(rgbImage);
title('Original image')
axis('on', 'image');
[R, G, B] = imsplit(rgbImage);
% Get mean of whole image
meanR = mean2(R)

meanR = 120.7808

meanG = mean2(G)

meanG = 66.3971

meanB = mean2(B)

meanB = 57.6335

rr=(double(R)-meanR).^2;
gg=(double(G)-meanG).^2;
bb=(double(B)-meanB).^2;
v=sqrt(rr+gg+bb);
subplot(2, 1, 2);
imshow(v, []);
title('Diff from mean')
colorbar

[DGM]

I added an example for that and several other possibilities.

[Maleeha Syed]

it works thank you!

[DGM]

That's the distance between all pixels and the image mean, not the mean of a set of selected pixels.

Answer 2

[DGM]

For RGB distance between a single pixel and the image mean:

% a small bit of a test image
inpict = imread('peppers.png');
inpict = inpict(1:10,1:10,:); % just pick a sample

% some points
x = 5;
y = 5;

meancolor = mean(inpict,1:2);
mypixel = double(inpict(y,x,:));

% distance between one point and mean color
Dpointmn = sqrt(sum((mypixel - meancolor).^2,3))

Dpointmn = 2.7740

For distance between all pixels and the image mean (see Dallmn(5,5)):

% distance between all points and mean color
Dallmn = sqrt(sum((double(inpict) - meancolor).^2,3))

Dallmn = 10×10

3.0356 1.8695 3.5937 2.8204 5.6511 0.5430 2.5090 1.5984 3.3697 4.2302 0.5430 3.0814 4.9634 2.4403 4.3675 3.7007 2.5090 1.1979 0.9460 2.2572 3.2886 1.0747 0.9460 2.9249 2.7740 2.3441 3.7088 2.6258 1.6837 2.1529 2.0037 4.0761 3.5909 2.2572 0.5430 1.8533 2.1389 4.5050 4.0786 4.9371 0.5430 2.0087 2.1389 2.1342 2.7740 2.4970 3.3727 2.7740 4.4827 3.0814 1.6837 6.6959 4.9573 3.4110 3.5348 4.1394 3.8672 1.5149 1.8533 3.3519 1.9378 2.9249 2.6636 3.6925 1.3019 3.5909 5.2759 2.5090 3.3429 5.5168 3.7168 3.8490 4.7429 2.8204 2.4525 3.2611 2.0087 1.5984 2.0087 2.5209 4.1899 1.3247 1.5984 1.6837 2.2748 2.7884 2.1529 2.9249 2.0087 4.1346 3.3786 1.7014 1.7014 3.8334 5.4913 2.5719 1.9378 1.8695 2.7740 2.1529

For the distance between all pixels and a single pixel:

% distance between all points and single color
Dall = sqrt(sum((double(inpict) - mypixel).^2,3))

Dall = 10×10

5.3852 4.2426 5.1962 1.7321 4.4721 3.1623 5.0990 3.1623 3.7417 3.6056 3.1623 5.8310 7.6811 2.4495 4.6904 6.0000 5.0990 2.4495 2.8284 3.1623 2.4495 3.3166 3.6056 1.0000 0 5.0000 6.4031 2.0000 2.2361 3.7417 3.7417 3.0000 2.2361 3.1623 2.2361 4.2426 3.7417 2.2361 3.3166 4.5826 3.1623 4.5826 3.0000 1.4142 0 3.3166 2.4495 0 4.1231 3.0000 3.1623 9.0554 7.3485 3.0000 2.2361 4.4721 4.1231 4.1231 4.2426 4.2426 4.5826 5.0990 4.1231 2.4495 2.0000 6.3246 7.8740 5.0990 2.8284 6.0828 4.5826 2.2361 2.2361 1.7321 2.4495 3.1623 2.4495 3.1623 4.5826 5.0990 6.4807 3.0000 3.1623 3.1623 3.4641 4.1231 4.3589 5.0990 4.5826 6.0828 5.3852 4.4721 4.4721 5.3852 6.0828 3.0000 2.4495 4.2426 5.0000 4.3589

For the distance between all pixels and the mean color local to a specific point:

% get a 3x3 local sample
r = 1; % sample radius
sx = min(max([x-r x+r],1),size(inpict,2));
sy = min(max([y-r y+r],1),size(inpict,1));
sample = inpict(sy(1):sy(2),sx(1):sx(2),:);

% find the local mean color
sampmean = mean(sample,1:2);

% distance between all points and local mean color
Dallmn = sqrt(sum((double(inpict) - sampmean).^2,3))

Dallmn = 10×10

4.3617 3.5935 4.8215 1.3833 4.5488 1.7708 3.3370 1.3426 2.0062 2.5652 1.7708 4.5731 6.1934 0.9558 3.3370 4.0307 3.3370 1.8922 2.4545 1.1653 2.1140 1.9213 2.6504 2.1915 2.0608 3.9892 4.7519 3.1662 1.3005 3.6549 2.4994 4.2586 3.8762 1.1653 1.4614 2.7532 2.8328 4.2586 4.7284 5.8044 1.7708 3.3866 2.6504 1.3426 2.0608 1.6741 2.4994 2.0608 5.3668 3.9892 1.6405 7.1198 5.6590 1.9783 1.8626 3.8038 4.9354 2.4771 2.7532 4.7167 3.2527 3.1309 2.0876 2.2167 0.4969 4.6930 6.8656 3.3370 3.8618 6.8575 4.8902 2.7330 3.3866 1.3833 2.4994 4.1663 2.9102 1.3426 3.3866 4.0582 5.8521 2.5652 1.3426 1.6405 3.3036 4.0991 3.7892 3.1309 3.3866 4.0445 3.5154 3.1662 3.1662 5.4490 6.7923 3.1837 2.7532 3.5935 3.0409 3.7892

To find the distance between all pixels and the mean color of a set of pixels defined by scattered points:

% some points (row vectors)
x = [4 5 6];
y = [5 5 5];

% get samples from selected locations
sz = size(inpict);
idx = sub2ind(sz(1:2),y,x) + [0;1;2]*prod(sz(1:2));
sample = permute(inpict(idx),[3 2 1]); % this is 1xNx3

% find mean color of selection
sampmean = mean(sample,1:2);

% distance between all points and mean color of selected pixels
Dallmn = sqrt(sum((double(inpict) - sampmean).^2,3))

Dallmn = 10×10

4.7140 3.7268 4.7842 1.2472 4.5338 2.1344 3.9016 1.9720 2.5604 2.7487 2.1344 4.9554 6.6500 1.3744 3.8152 4.6786 3.9016 1.7951 2.4267 1.7951 2.1344 2.3570 2.8674 1.6997 1.3744 4.2687 5.3125 2.6874 1.2472 3.5434 2.9250 3.8586 3.2998 1.7951 1.4907 3.1447 2.9250 3.5901 4.1899 5.3125 2.1344 3.6818 2.4944 1.1055 1.3744 2.0548 2.1344 1.3744 4.9216 3.5901 1.9720 7.7388 6.1554 2.0548 1.6997 4.0689 4.6428 2.9814 3.1447 4.4597 3.6818 3.8152 2.7487 1.9720 0.7454 5.2175 7.1570 3.9016 3.3500 6.5490 4.7842 2.4944 2.8674 1.2472 2.4267 3.7268 2.5604 1.9720 3.6818 4.3843 6.0461 2.6247 1.9720 1.9720 3.1447 3.9441 3.8586 3.8152 3.6818 4.7140 4.1899 3.5434 3.5434 5.3748 6.5490 3.0912 2.5604 3.7268 3.6818 3.8586

Note that these examples will work regardless of how many channels the image has. No need to split or handle RGB images differently than gray images.

At least for 1976 Delta E, this is the same idea that you can apply in LAB (if you don't have deltaE)