My custom implementation of MATLAB's imdilate

I have a few different comments.

Firstly, for 2D arrays like in your example, you can simply do conv2(binaryImage,structuringElement,'same'), it's equivalent to imdilate (and about 4 times faster). That might not be acceptable for your case if you are supposed to write the code yourself though.

Next, and importantly, your code has a logic error. If your binary image has a white pixel along any edge, the results are incorrect or you get an error (if it's in the top row). This is due to your indexing into dilatedImage where i+k-2 can be less than 1, or can be larger than the number of rows in dilatedImage.

You have a lot of comments, which is better than none, but I think it's excessive here. It is very clear what this line of code does [imageRows, imageColumns] = size(binaryImage);, for example, even for people who don't know Matlab. Some of your comments make the lines very long as well.

A minor issue, I'd move the imdilate command somewhere else, that part of the code is where you are re-creating imdilate, I think it's nicer to separate the verification part.

Another minor issue is in using i and j as loop indices. Since i and j can be used for complex numbers in Matlab, by convention they are avoided as loop indices (although I do not usually worry about this).

I would put the code that does the work into its own function that takes binaryImage and structuringElement as inputs and returns dilatedImage.

The big thing here is you 6 nested for/if statements, which is hard to read and hard to code. Historically, such nested loops would have been very slow in Matlab, but nowadays it's not such a concern. One thing you should do is swap the order of the first two for loops. Matlab stores arrays in a column major format, so loop over the columns first, then the rows. This gives roughly a 20-25% speedup in my tests. Additionally, checking to remove the out-of-bounds error is also a speedup, since we only set pixels to one if they are inside the image, irregardless of the value of the structuring array. Check out this version of your loop:

for j = 1:imageColumns
    for i = 1:imageRows
        if binaryImage(i, j) == 1
            % Loop through the entries of the structuring element
            for l = 1:structuringColumns
                for k = 1:structuringRows
                    % make sure the new pixel is inside the image
                    if  i+k-2>0 && j+l-2>0 && i+k-2<=imageRows && j+l-2<=imageColumns
                        % make a white pixel if necessary
                        if structuringElement(k, l)==1
                            dilatedImage(i + k - 2, j + l - 2) = 1;
                        end
                    end
                end
            end
        end
    end
end

Of course it would be nice to avoid writing all these loops, and that can be done. This next method is only slightly faster, but to me is cleaner. The idea is to first find the white pixels, and just loop over them, for each one working out which surrounding pixels should be changed.

[rowIdx,colIdx] = find(binaryImage); % the indices of white pixels in the image
[maskRowIdx,maskColIdx] = find(structuringElement); % indices of structuring elements

% now get indices of structuring elements relative to the centre of the array
maskRowIdx = maskRowIdx - floor(structuringRows/2) - 1;
maskColIdx = maskColIdx - floor(structuringColumns/2) - 1;

dilatedImage = zeros(imageRows, imageColumns);

% loop over each white pixel in the image
for i = 1:numel(rowIdx)
    % these are just the indices of the pixel
    rI = rowIdx(i);
    cI = colIdx(i);

    % now loop over each non-zero element of the structuring array
    for j = 1:numel(maskRowIdx)
        % the position of the pixel to change is (r,c)
        r = rI+maskRowIdx(j);
        c = cI+maskColIdx(j);

        % if the pixel is inside the image, we make it a 1
        if r>0 && c>0 && r<=imageRows && c<=imageColumns
            dilatedImage(r,c) = 1;
        end
    end
end