Segmentation & confusion matrix for colored image; matrix dimensions must agree.

Pls I'm having issues getting my confusion matrix done. Below are all code-files and picture '1.jpg' is NOT a monochrome image. On plotting the confusion matrix, it keeps giving me

'matrix dimensions must agree'

thank u for the quick response.

[Sreeja Banerjee]

I see that your code requires the file '1.jpg' in order to execute. Could you please attach the file so that I can try to reproduce the error?

[bayoishola20]

okay! Thank you so much, here is the file

Answer 1

[Ping-Chang Shih]

Given that the input image is not a monochrome image, all elements in your "segmented_images" cell array have three dimensions. However, the polygonal ROI image you create (stored as variable "BW_1") has two dimensions. Plugging the two types of variables into the MATLAB "confusion" function causes an error since the two arguments are expected to have the same number of dimensions.

The above conclusion is obtained from my interpretation of your code. If I misunderstood your code, please let me know.

[bayoishola20]

You are just exactly right. I now need to fix the dimension issue and plot the confusion matrix, which I have not been successful at so far. I would appreciate your help.

Here is the image. http://www.mathworks.com/matlabcentral/answers/uploaded_files/18855/1.jpg

[Image Analyst]

Are you still having trouble? If so, post your latest code.

[bayoishola20]

Yes I am.

For image segmentation.

he= imread('C:\Users\ISHOLA\Desktop\1.jpg');
imshow(he), title('ORIGINAL IMAGE');
cform = makecform('srgb2lab');
lab_he = applycform(he,cform);
ab = double(lab_he(:,:,2:3));
nrows = size(ab,1);
ncols = size(ab,2);
disp(nrows);
disp(ncols);
ab = reshape(ab,nrows*ncols,2);
nColors = 6;

[cluster_idx cluster_center] = kmeans(ab,nColors,'distance','sqEuclidean','replicates',6);
pixel_labels = reshape(cluster_idx,nrows,ncols);
%figure, imshow(pixel_labels,[]), title('image labeled by cluster index');
segmented_images = cell(1,3);
rgb_label = repmat(pixel_labels,[1 3]);

for k = 1:nColors
    color = he;
    color(rgb_label ~= k) = 0;
    segmented_images{k} = color;
end

For dataset training;

%training of classes
imshow(he);
%Training of class 1
BW_1=roipoly(he);
%Training of class 2
BW_2=roipoly(he);
%Training of class 3
BW_3=roipoly(he);
%Training of class 4
BW_4=roipoly(he);
%Training of class 5
BW_5=roipoly(he);
%Training of class 6
BW_6=roipoly(he);

On the aspect of the confusion Matrix, I don't really know how to put it although I know that "pixel_labels" is what I need from the segmentation done. I tried this for instance and it tells me

'Index exceeds matrix dimensions.'

confusionMatrix(r,c)=confusionMatrix(pixel_labels==1,BW_1) +1;

But how to get it right to plot the confusion matrix is what I don't still understand. Thank you.

Thanks for the help on this link though still need your help. http://www.mathworks.com/matlabcentral/answers/157393-kmeans-image-segmentation-and-confusion-matrix

Answer 2

[Nurudeen Ayanleye Yusuff]

please i need code for ISODATA which i will implement in my GUI i created. Also i need how to code accuracy assessment for classified image using kmean of unsupervised classification