Can I use grayscale image in Lab color space

Hello, i tried one image processing tutorial, which i found on YouTube. Basically it's about segmentation of salami pieces on a pizza. For that i have to convert an RGB-image into the CIELAB color space and define different classes. This code works perfectly fine.

Now my question is: Can I use this script to segment things from a gray-scale image? Or do I have to convert the grayscale image into the RGB format?

Here the code:

% Load picture
img = imread('Pizza.jpg');

% Show original pic
subplot(4,3,1);
imshow(img);

% Convert into lab
labimg = rgb2lab(img);

% Adjust and show color channels
subplot(4,3,2);
imshow(labimg(:,:,1),[]);
subplot(4,3,3);
imshow(labimg(:,:,2),[]);
subplot(4,3,4);
imshow(labimg(:,:,3),[]);

% Define classes
salami = labimg(508:603, 200:305, :);
tomato = labimg(634:729, 326:457, :);
plate  = labimg(645:708, 615:699, :);
backgr = labimg(3:56, 5:53, :);

% Save averages from classes in a matrix
classes = [[mean2(salami(:,:,2)), mean2(salami(:,:,3))]', [mean2(tomato(:,:,2)), mean2(tomato(:,:,3))]', [mean2(plate(:,:,2)), mean2(plate(:,:,3))]', [mean2(backgr(:,:,2)), mean2(backgr(:,:,3))]']; 

[height, width, channels] = size(labimg);

classimg = zeros(height,width);

for i = 1:height
    for j = 1:width
        classimg(i,j) = nearestNeighbour2d(labimg(i,j,2:3), classes);
    end
end

% Show class-image 
subplot(4,3,5);
imshow(classimg, []);

% Extraction of salami
salamiclass = (classimg ==1);
salamiclass = double(salamiclass);

subplot(4,3,6);
imshow(salamiclass, []);

% Use erode
se = strel('disk',28);
erodedBW = imerode(salamiclass, se);

subplot(4,3,7);
imshow(erodedBW, []);

% Extraction of the regioons
labels = bwlabel(erodedBW);
subplot(4,3,8);
imshow(labels, []);

numLabels = max(max(labels));

regions = zeros(5,numLabels);
for ii = 1:numLabels
    [xx, yy] = find(labels==ii);
    regions(1:5,ii) = [min(yy) min(xx) max(yy) max(xx) size(xx,1)]';
end

circles = zeros(numLabels,3);

for ii=1:numLabels
    circles(ii,1:3) = [regions(3,ii)-(regions(3,ii)-regions(1,ii))/2 regions(4,ii)-(regions(4,ii)-regions(2,ii))/2 max([regions(3,ii)-regions(1,ii) regions(4,ii)-regions(2,ii)])];
end

img = draw('Circles', img, circles, [0 0 255], 5);

subplot(4,3,9);
imshow(img, []);

[Stephen23]

S.M.'s "Answer" moved here:

Nobody an idea?

Answer 1

[Walter Roberson]

In theory it would be possible to calculate the L component directly knowing the grayscale value and making assumptions about the whitepoint . The a and b components would both be 0 for true grayscale. You would use that code to replace the labimg = rgb2lab(img); call.

In practice, it is far easier just to code

labimg = rgb2lab( repmat(img, [1 1 3]) );

[S.M.]

Thank you Walter, I'm much further now. I'll try it with another picture. Let's see what happens. Thanks again!