how to pass image to function after k-means?

hi folks,

I want to perform a watershed transform on an image after performing a k-means clustering, below is my code.

img = imread('myimage.jpg');

lab_img = rgb2lab(img);

ab = lab_img(:,:,2:3);
a = lab_img(:,:,2);
b = lab_img(:,:,3);
ab = im2single(ab);
nColours = 2;
k = 20;
pixelLabels = imsegkmeans(ab, nColours, "NumAttempts", k);

imshow(pixelLabels,[])
title('Image Labeled by Cluster Index');
 parfor i = 1 : k
     maskID = strcat('mask', i);
     clusterID = strcat('cluster', i);
     maskID = pixelLabels == i
     clusterID = img.*uint8(maskID);
     figure(i), imshow(clusterID)
     title(['Objects in Cluster ', num2str(i)]);
 end

My question is: how can I pass the result of the k-means to the watershed function? I tried to pass "clusterID" and "cluster20" but it returned an error.

Any help would be appreciated, thanks in advance!

[Adam]

What was the error? And what is "cluster20"?

[Teshan Rezel]

@Adam, "cluster20" is the value of clusterID after the final iteration of the for loop, which is set to 20.

The error I get is that it is not a recognised variable.

[Adam]

Well, it's a parfor loop so the final iteration is not guaranteed to be k = 20.

I can't remember off the top of my head, but if you declare clusterID before the parfoor loop then it may still exist after it, though I doubt in a satisfactory way if it does, as there is no guarantee which result will complete last and I suspect you would get a warning or error on the variable.

Simplest option would be to store the results in a 3d array so you have all of them after the parfor loop.

Answer 1

[Image Analyst]

There is a watershed() function. Use it.

Example of watershed: Image Processing Blog

[Teshan Rezel]

I've tried, but as my question stated, the problem was less about the watershed function itself and more about passing an argument to it. Thanks for your response though