Label array from workspace in Image Labeler

I would like to label the entries of an array Z using MATLAB's imageLabeler. Specifically, Z is an MxN double matrix that I can visualize as an image by calling imagesc(Z). As an end goal, I would like an MxN array G such that G(i,j) = k indicates that the (i,j)-entry of Z is class k.

My main questions are as follows:

1. How can I format Z to be suitable for loading into imageLabeler? I understand that imageLabeler can be applied to any image that is loaded using imread, but I'm not sure how to apply it to a double array.
2. Once I've obtained my ground truth labels using imageLabeler, how can I convert the groundTruth object into an array G of the form described above?

Thank you very much for your help.

Answer 1

[recent works]

To use MATLAB's Image Labeler with a matrix Z, you can convert it into a grayscale image and then load it into the Image Labeler tool. Afterwards, you can convert the labeled regions back into an array G with the corresponding class labels. Here's a step-by-step guide:

1. Format Z for Image Labeler: Convert your Z matrix into a grayscale image suitable for Image Labeler. You can achieve this by scaling Z to the range [0, 255] and converting it to uint8 format.

Z_scaled = mat2gray(Z) * 255; % Scale Z to [0, 255]

Z_uint8 = uint8(Z_scaled); % Convert to uint8

Now, you can save this as an image file and load it into the Image Labeler tool. You can use imwrite to save it as a PNG, JPEG, or any other format supported by Image Labeler.

imwrite(Z_uint8, 'Z_image.png'); % Save as an image file

2. Labeling with Image Labeler: Open MATLAB's Image Labeler tool and load the image file Z_image.png you just created. You can then label the regions in the image using the tool.

3. Convert Ground Truth Labels to Array G: Once you've labeled the regions in Image Labeler, you can export the labeled data as a groundTruth object in MATLAB. Then, you can convert this object into an array G with the desired format.

% Export ground truth labels from Image Labeler

groundTruthData = groundTruth('Z_image.png', labels);

% Convert to array G

G = zeros(size(Z)); % Initialize G with zeros

for i = 1:numel(groundTruthData.LabelData)

labelData = groundTruthData.LabelData(i);

label = labelData{1}; % Extract label

region = labelData{2}; % Extract region coordinates

G(region) = label; % Assign label to corresponding region in G

end

In this code, labels should be a cell array containing the labels assigned to each region in Image Labeler.

Now, G will be an array where G(i,j) = k indicates that the (i,j) entry of Z is in class k, as you desired.