Image ROI Label Issue

Hello,

I currently have a code which outputs images, I would like to label different ROI within these images using consecutive numbers. I have written some code to compute the centroid of the ROIs within the image that I would like to label, this data is stored in "props".

"props" is a 1 x 4 cell (I have 4 images, each cell corresponds to a different image). Each of the cells in "props" contains a 5 (the current images have 5 different ROI each) x 1 struct with 2 fields (Centroid Position and Mean Intensity). Here is the code I have thus far:

 for i = i:length(props)
    imagesc(boutput{1,i});
    axis image
    colormap('gray')
 end

labelShiftX = -7; %simply for aligning number placement
for k = length(props{1})
   cent = props{k}.Centroid;
   text(cent(1) + labelShiftX, cent(2), num2str(k));
end

To note, "props" was generated using the regionprops function which takes in boutput and output, the binarized and original images, respectively.

So far, my code is not accomplishing the task and the ROIs within the image are not labeled. Any suggestions on how to proceed?

Thank you for your time and attention!!!!

[Image Analyst]

If props was generated by regionprops(), it's a structure array, not a cell array, and you'd use parentheses, not braces. Or it could be a table if you asked for a table. Please show the entire code.

[itend]

Here is the bit of code that generates props: props = cell(size(output,1),1);

    for i = 1:size(output,2)
props{i} = regionprops(labelboutput{1,i},output{1,i},'MeanIntensity','Centroid'); 
    end

What other parts of the code would you like to see? Thanks!

Answer 1

[Image Analyst]

I think you need to extract out the props first. Try this:

thisProps = props{imageIndex}; % Extract props for this image only.
cent = thisProps(blobIndex).Centroid;

where imageIndex is the index (number) of the image you extracted from, and blobIndex is the index of the blob within that particular image. Be careful to use the braces and parentheses just like I did. So you need to decide which image you're working on and then within that image, what blob you're working on.

[itend]

I have modified my code to reflect your suggestion:

    for k = 1:length(props)
        thisProps = props{k};
        cent = thisProps(k).Centroid;
        text(cent(1) + labelShiftX, cent(2), num2str(k)); 
    end

This successfully labels the ROI in only the last of images. Any ideas on how to modify this code to get the ROIs in every image labeled?

Thank you :)

[Image Analyst]

Of course that's what it does, because you didn't do what I said. You're using k for both. Why? I didn't tell you to do that. I showed you how there are two different indexes, blobIndex over blobs within a particular image, and imageIndex over all images. For some reason you chose to ignore my advice and have only one loop instead of two and you have only one index, k, instead of 2. Do what I say and it should display all images with each blob in each index with a text label over it.

[itend]

Is this what you meant? It didn't seem to resolve the issue :/

    for k = 1:length(props)
        thisProps = props{k};
        for i = 1:length(props)
            cent = thisProps(k).Centroid;
            text(cent(1) + labelShiftX, cent(2), num2str(i)); 
        end
    end

I apologize if I am misunderstanding what you are saying, I am a novice at matlab.

Thank you for taking the time to help me!

[Image Analyst]

Almost, but there is no guarantee that the number of blobs will equal the number of images. So you need to have the inner loop go to the number of blobs and change the index from k to i:

for k = 1:length(props)
        thisProps = props{k};
        for i = 1:length(thisProps)
            cent = thisProps(i).Centroid;
            text(cent(1) + labelShiftX, cent(2), num2str(i)); 
        end
end

However, they will all be displayed on the currently displayed image, while the blobs were found on different images. So after you get thisProps you might want to display the k'th image, perhaps by calling figure() or subplot() and then imshow().

[itend]

Huzzah! That did the trick.

Like always, thank you so much for your advice and expertise!