How to separate a set of pixels

Hello,

After filtering my image (attached) that contains a large bubble, I arrived at image 5. My goal is to measure, even with low precision, the position of the tip of the larger bubble. Could you suggest any alternative so that I can separate the first set of white pixels from the bottom of the others? My goal is to exclude any pixel set other than the bottom one in image 5 so i can measure the position of the tip of the larger bubble.

Thanks in advance !

image = imread("image.jpg");
cropped = imcrop(image,[41 0 382 1024]);
figure, imshow(cropped);

gray = rgb2gray(cropped);
figure, imshow(gray);

threhsold = im2bw(gray, 0.5);
figure, imshow(threhsold);

remove = bwareaopen(threhsold,3500);
figure, imshow(remove);

se = strel('line',410,0);
closing = imclose(remove,se);
figure, imshow(closing);

originalLine = closing(1 , :);
originalLine2 = closing(end , :);
closing(1, :) = true;
closing(end, :) = true;
filtered = imfill(closing, 'holes');
closing(end,:) = false;
closing(1,:) = false;
filtered(1, :) = originalLine;
filtered(end, :) = originalLine2;
figure, imshow(filtered);

props = regionprops(filtered, 'BoundingBox');
for k = 1 : length(props)
    thisBB = props(k).BoundingBox;
    y(k) = thisBB(2) + thisBB(4);
end
[~, indexOfLowest] = max(y);
% Label the image
labeledImage = bwlabel(filtered);
% Extract the lowest blob
lowestBlob = ismember(labeledImage, indexOfLowest);
figure, imshow(lowestBlob);

Answer 1

[Image Analyst]

You can get the bounding box of all of them, then find the one with the lowest bottom edge. Something like (untested)

props = regionprops(mask, 'BoundingBox');
for k = 1 : length(props)
    thisBB = props(k).BoundingBox;
    y(k) = thisBB(2) + thisBB(4);
end
[~, indexOfLowest] = max(y);
% Label the image
labeledImage = bwlabel(mask);
% Extract the lowest blob
lowestBlob = ismember(labeledImage, indexOfLowest);

Or you can use find() to find the row and column of the lowest pixel and use imreconstruct() to extract the blob that contains that pixel. Something like (untested)

[r, c] = find(mask); % Get coordinates of all white pixels.
% Find lowest (max r)
[rMax, indexOfLowest] = max(r)
% Make a marker image
markerImage = false(size(mask));
% Set that one pixel to true
markerImage(rMax, c(indexOfLowest)) = true;
% Use imreconstruct() to extract the lowest blob.
lowestBlob = imreconstruct(markerImage, mask);

[Jórdan Venâncio Leite]

Image Analyst, the mask is not working with my code, could you please help me with this?

I've attached the code above

[Image Analyst]

You forgot to attach the code. The only code I see from you is in your original post and it looks like it is using neither suggestion from me. I don't see ismember() or imreconstruct(), etc. So you forgot to attach the code where you tried my suggestion.

If you simply want the row of the lowest part of any bubble in your mask you can do this:

[r, c] = find(filtered); % Get coordinates of all white pixels.
% Find lowest row (max r).  Lowest row is rMax.
[rMax, indexOfLowest] = max(r)

[Jórdan Venâncio Leite]

Thanks for you answer Image Analyst! I've uploaded the code now, sorry. I have one problem now:

in image 11 that is attached there is no big bubble. So, I would like the answer for this image to be 0, so I would be storing in the position vector of the tips of the big bubbles that in this image there is no big bubble, so, position 0

[Image Analyst]

props = regionprops(filtered, 'BoundingBox');
if length(props) == 0
    warningMessage = 'No bubbles were found in this image';
    uiwait(warndlg(warningMessage));
    return
end

[Jórdan Venâncio Leite]

I did a test with this code above using the image '11' and the value of lenght 'props' is always different from 0 and it ended up not working

image = imread("11.jpg");
cropped = imcrop(image,[41 0 382 1024]);
figure, imshow(cropped);

gray = rgb2gray(cropped);
figure, imshow(gray);

threhsold = im2bw(gray, 0.5);
figure, imshow(threhsold);

remove = bwareaopen(threhsold,3500);
figure, imshow(remove);

se = strel('line',410,0);
closing = imclose(remove,se);
figure, imshow(closing);

originalLine = closing(1 , :);
originalLine2 = closing(end , :);
closing(1, :) = true;
closing(end, :) = true;
filtered = imfill(closing, 'holes');
closing(end,:) = false;
closing(1,:) = false;
filtered(1, :) = originalLine;
filtered(end, :) = originalLine2;
figure, imshow(filtered);

props = regionprops(filtered, 'BoundingBox');
if length(props) == 0
    warningMessage = 'No bubbles were found in this image';
    uiwait(warndlg(warningMessage));
    return
end

[Image Analyst]

You don't need regionprops(). Like I said, to find the last line of the lowest blob, simply do this:

[r, c] = find(filtered); % Get coordinates of all white pixels.
% Find lowest row (max r).  Lowest row is rMax.
[rMax, indexOfLowest] = max(r)
yline(rMax, 'Color', 'r', 'LineWidth', 2)

[yanqi liu]

yes，sir，may be use watershed method to get image segment

[Jórdan Venâncio Leite]

Thanks for your answer yanqi liu