kmeans Segmentation Clarification for color images

I had segmented the attached image using kmeans. After kmeans color image segmentation, when I had plotted the cluster centers, I found that, the cluster center was not at the correct point. Why?

[Walter Roberson]

Please show your code.

[Sabarinathan Vadivelu]

dataImage = imread('fpot3_1.jpg');
cform = makecform('srgb2lab');
lab_he = applycform(dataImage,cform);
ab = double(lab_he(:,:,2:3));
nrows = size(ab,1);
ncols = size(ab,2);
ab = reshape(ab,nrows*ncols,2);
nColors = 3;
[cluster_idx,cluster_center] = kmeans(ab,nColors,'emptyaction','singleton');
pixel_labels = reshape(cluster_idx,nrows,ncols);
imshow(pixel_labels,[]); hold on;
plot(cluster_center(:,2),cluster_center(:,1),'r*')

[Nishandhini Balasundar]

can you please explain each line of k means segmentation?

[Image Analyst]

You can try mine (attached). I always put in lots of comments.

Answer 1

[Anand]

Your cluster centers are not in the XY co-ordinate space. They are in the AB color space. So cluster_center does not refer to a location on the image.

Answer 2

[Image Analyst]

I think perhaps this demo is what you're after: http://www.mathworks.com/products/demos/image/color_seg_k/ipexhistology.html

[Image Analyst]

You might like the color segmentation demos in my File Exchange, or this:

http://www.mathworks.com/matlabcentral/fileexchange/37197-dem-diffused-expectation-maximisation-for-image-segmentation

[Sabarinathan Vadivelu]

I closely following up your file exchange list. I had worked with all your demos in that list. But right now I'm working with kmeans Color Image Segmentation.

[Anand]

What didn't work in the link Image Analyst suggested?